3 Questions: Leveraging carbon uptake to lower concrete’s carbon footprint

To secure a more sustainable and resilient future, we must take a careful look at the life cycle impacts of humanity’s most-produced building material: concrete. Carbon uptake, the process by which cement-based products sequester carbon dioxide, is key to this understanding.

Hessam AzariJafari, the MIT Concrete Sustainability Hub’s deputy director, is deeply invested in the study of this process and its acceleration, where prudent. Here, he describes how carbon uptake is a key lever to reach a carbon-neutral concrete industry.

Q: What is carbon uptake in cement-based products and how can it influence their properties?

A: Carbon uptake, or carbonation, is a natural process of permanently sequestering CO₂ from the atmosphere by hardened cement-based products like concretes and mortars. Through this reaction, these products form different kinds of limes or calcium carbonates. This uptake occurs slowly but significantly during two phases of the life cycle of cement-based products: the use phase and the end-of-life phase.

In general, carbon uptake increases the compressive strength of cement-based products as it can densify the paste. At the same time, carbon uptake can impact the corrosion resistance of concrete. In concrete that is reinforced with steel, the corrosion process can be initiated if the carbonation happens extensively (e.g., the whole of the concrete cover is carbonated) and intensively (e.g., a significant proportion of the hardened cement product is carbonated). [Concrete cover is the layer distance between the surface of reinforcement and the outer surface of the concrete.]

Q: What are the factors that influence carbon uptake?

A: The intensity of carbon uptake depends on four major factors: the climate, the types and properties of cement-based products used, the composition of binders (cement type) used, and the geometry and exposure condition of the structure.

In regard to climate, the humidity and temperature affect the carbon uptake rate. In very low or very high humidity conditions, the carbon uptake process is slowed. High temperatures speed the process. The local atmosphere’s carbon dioxide concentration can affect the carbon uptake rate. For example, in urban areas, carbon uptake is an order of magnitude faster than in suburban areas.

The types and properties of cement-based products have a large influence on the rate of carbon uptake. For example, mortar (consisting of water, cement, and fine aggregates) carbonates two to four times faster than concrete (consisting of water, cement, and coarse and fine aggregates) because of its more porous structure.The carbon uptake rate of dry-cast concrete masonry units is higher than wet-cast for the same reason. In structural concrete, the process is made slower as mechanical properties are improved and the density of the hardened products’ structure increases.

Lastly, a structure’s surface area-to-volume ratio and exposure to air and water can have ramifications for its rate of carbonation. When cement-based products are covered, carbonation may be slowed or stopped. Concrete that is exposed to fresh air while being sheltered from rain can have a larger carbon uptake compared to cement-based products that are painted or carpeted. Additionally, cement-based elements with large surface areas, like thin concrete structures or mortar layers, allow uptake to progress more extensively.

Q: What is the role of carbon uptake in the carbon neutrality of concrete, and how should architects and engineers account for it when designing for specific applications?

A: Carbon uptake is a part of the life cycle of any cement-based products that should be accounted for in carbon footprint calculations. Our evaluation shows the U.S. pavement network can sequester 5.8 million metric tons of CO₂, of which 52 percent will be sequestered when the demolished concrete is stockpiled at its end of life.

From one concrete structure to another, the percentage of emissions sequestered may vary. For instance, concrete bridges tend to have a lower percentage versus buildings constructed with concrete masonry. In any case, carbon uptake can influence the life cycle environmental performance of concrete.

At the MIT Concrete Sustainability Hub, we have developed a calculator to enable construction stakeholders to estimate the carbon uptake of concrete structures during their use and end-of-life phases.

Looking toward the future, carbon uptake’s role in the carbon neutralization of cement-based products could grow in importance. While caution should be taken in regards to uptake when reinforcing steel is embedded in concrete, there are opportunities for different stakeholders to augment carbon uptake in different cement-based products.

Architects can influence the shape of concrete elements to increase the surface area-to-volume ratio (e.g., making “waffle” patterns on slabs and walls, or having several thin towers instead of fewer large ones on an apartment complex). Concrete manufacturers can adjust the binder type and quantity while delivering concrete that meets performance requirements. Finally, industrial ecologists and life-cycle assessment practitioners need to work on the tools and add-ons to make sure the impact of carbon is well captured when assessing the potential impacts of cement-based products in buildings and infrastructure systems.

Currently, the cement and concrete industry is working with tech companies as well as local, state, and federal governments to lower and subsidize the code of carbon capture sequestration and neutralization. Accelerating carbon uptake where reasonable could be an additional lever to neutralize the carbon emissions of the concrete value chain.

Carbon uptake is one more piece of the puzzle that makes concrete a sustainable choice for building in many applications. The sustainability and resilience of the future built environment lean on the use of concrete. There is still much work to be done to truly build sustainably, and understanding carbon uptake is an important place to begin.

de MIT News https://ift.tt/b9pP1t0

Speeding up drug discovery with diffusion generative models

With the release of platforms like DALL-E 2 and Midjourney, diffusion generative models have achieved mainstream popularity, owing to their ability to generate a series of absurd, breathtaking, and often meme-worthy images from text prompts like “teddy bears working on new AI research on the moon in the 1980s.” But a team of researchers at MIT's Abdul Latif Jameel Clinic for Machine Learning in Health (Jameel Clinic) thinks there could be more to diffusion generative models than just creating surreal images — they could accelerate the development of new drugs and reduce the likelihood of adverse side effects.

A paper introducing this new molecular docking model, called DiffDock, will be presented at the 11th International Conference on Learning Representations. The model's unique approach to computational drug design is a paradigm shift from current state-of-the-art tools that most pharmaceutical companies use, presenting a major opportunity for an overhaul of the traditional drug development pipeline.

Drugs typically function by interacting with the proteins that make up our bodies, or proteins of bacteria and viruses. Molecular docking was developed to gain insight into these interactions by predicting the atomic 3D coordinates with which a ligand (i.e., drug molecule) and protein could bind together. 

While molecular docking has led to the successful identification of drugs that now treat HIV and cancer, with each drug averaging a decade of development time and 90 percent of drug candidates failing costly clinical trials (most studies estimate average drug development costs to be around $1 billion to over $2 billion per drug), it’s no wonder that researchers are looking for faster, more efficient ways to sift through potential drug molecules.

Currently, most molecular docking tools used for in-silico drug design take a “sampling and scoring” approach, searching for a ligand “pose” that best fits the protein pocket. This time-consuming process evaluates a large number of different poses, then scores them based on how well the ligand binds to the protein.

In previous deep-learning solutions, molecular docking is treated as a regression problem. In other words, “it assumes that you have a single target that you’re trying to optimize for and there’s a single right answer,” says Gabriele Corso, co-author and second-year MIT PhD student in electrical engineering and computer science who is an affiliate of the MIT Computer Sciences and Artificial Intelligence Laboratory (CSAIL). “With generative modeling, you assume that there is a distribution of possible answers — this is critical in the presence of uncertainty.”

“Instead of a single prediction as previously, you now allow multiple poses to be predicted, and each one with a different probability,” adds Hannes Stärk, co-author and first-year MIT PhD student in electrical engineering and computer science who is an affiliate of the MIT Computer Sciences and Artificial Intelligence Laboratory (CSAIL). As a result, the model doesn't need to compromise in attempting to arrive at a single conclusion, which can be a recipe for failure.

To understand how diffusion generative models work, it is helpful to explain them based on image-generating diffusion models. Here, diffusion models gradually add random noise to a 2D image through a series of steps, destroying the data in the image until it becomes nothing but grainy static. A neural network is then trained to recover the original image by reversing this noising process. The model can then generate new data by starting from a random configuration and iteratively removing the noise.

In the case of DiffDock, after being trained on a variety of ligand and protein poses, the model is able to successfully identify multiple binding sites on proteins that it has never encountered before. Instead of generating new image data, it generates new 3D coordinates that help the ligand find potential angles that would allow it to fit into the protein pocket.

This “blind docking” approach creates new opportunities to take advantage of AlphaFold 2 (2020), DeepMind’s famous protein folding AI model. Since AlphaFold 1’s initial release in 2018, there has been a great deal of excitement in the research community over the potential of AlphaFold’s computationally folded protein structures to help identify new drug mechanisms of action. But state-of-the-art molecular docking tools have yet to demonstrate that their performance in binding ligands to computationally predicted structures is any better than random chance.

Not only is DiffDock significantly more accurate than previous approaches to traditional docking benchmarks, thanks to its ability to reason at a higher scale and implicitly model some of the protein flexibility, DiffDock maintains high performance, even as other docking models begin to fail. In the more realistic scenario involving the use of computationally generated unbound protein structures, DiffDock places 22 percent of its predictions within 2 angstroms (widely considered to be the threshold for an accurate pose, 1Å corresponds to one over 10 billion meters), more than double other docking models barely hovering over 10 percent for some and dropping as low as 1.7 percent.

These improvements create a new landscape of opportunities for biological research and drug discovery. For instance, many drugs are found via a process known as phenotypic screening, in which researchers observe the effects of a given drug on a disease without knowing which proteins the drug is acting upon. Discovering the mechanism of action of the drug is then critical to understanding how the drug can be improved and its potential side effects. This process, known as “reverse screening,” can be extremely challenging and costly, but a combination of protein folding techniques and DiffDock may allow performing a large part of the process in silico, allowing potential “off-target” side effects to be identified early on before clinical trials take place.

“DiffDock makes drug target identification much more possible. Before, one had to do laborious and costly experiments (months to years) with each protein to define the drug docking. But now, one can screen many proteins and do the triaging virtually in a day,” Tim Peterson, an assistant professor at the University of Washington St. Louis School of Medicine, says. Peterson used DiffDock to characterize the mechanism of action of a novel drug candidate treating aging-related diseases in a recent paper. “There is a very ‘fate loves irony’ aspect that Eroom’s law — that drug discovery takes longer and costs more money each year — is being solved by its namesake Moore’s law — that computers get faster and cheaper each year — using tools such as DiffDock.”

This work was conducted by MIT PhD students Gabriele Corso, Hannes Stärk, and Bowen Jing, and their advisors, Professor Regina Barzilay and Professor Tommi Jaakkola, and was supported by the Machine Learning for Pharmaceutical Discovery and Synthesis consortium, the Jameel Clinic, the DTRA Discovery of Medical Countermeasures Against New and Emerging Threats program, the DARPA Accelerated Molecular Discovery program, the Sanofi Computational Antibody Design grant, and a Department of Energy Computational Science Graduate Fellowship.

de MIT News https://ift.tt/cVSzNTe

A shot in the arm

Biologics, a class of therapeutics derived from living organisms, offer enormous advantages to patients battling challenging diseases and disorders. Treatments based on biologics can boost the immune system to stem attacks from infections or target specific pathways to block the formation of tumors.

“These drugs, which have been around for just the last 20 years, do magic,” says Amir Erfani, a postdoc in the MIT Department of Chemical Engineering (ChemE). “They can save millions of people around the world.”

But the unrivaled effectiveness of biologics comes at a cost. They can be difficult to administer, often demanding time-consuming intravenous (IV) infusions at clinics. Whether for patients struggling with life-threatening or lifelong conditions, the prospect of spending hours away from home, every few weeks, can prove extremely daunting.

Now, new work from an MIT team in collaboration with the Merck pharmaceutical company, which funded the research, suggests a practical solution for surmounting the difficulty of administering biologics. In a recent paper published in Advanced Healthcare Materials, these researchers describe a hydrogel platform for delivering monoclonal antibodies (MABs) — one type of biologic — through subcutaneous injection.

Erfani is lead author of the paper. Co-authors include Jeremy M. Schieferstein, a postdoc in ChemE at the time of the study, now senior scientist at Elektrofi; Apoorv Shanker, a postdoc in ChemE; Paula Hammond, Institute Professor and head of ChemE; and Patrick S. Doyle, the Robert T. Haslam (1911) Professor of Chemical Engineering, as well as Merck researchers.

“This is an important milestone,” says Doyle. “We are on the route to transforming the next generation of treatment with monoclonal antibodies and other kinds of therapeutics.”

Higher-test antibodies

Unlike most conventional drugs that are formulated chemically and comprised of small molecules, biologics are sprawling and unruly chains of proteins, sugars, and DNA segments, genetically engineered from living sources. These giant organic molecules don’t lend themselves to the kind of neat, dense packaging typically found in synthesized pills or injectable suspensions.

Take the MAB on which Erfani and Doyle focused, called pembrolizumab, or pembro for short. This unique antibody binds to a receptor associated with mediating immune responses to tumor cells, and is used in a range of sometimes intractable cancers. Pembro is normally administered in a dilute solution by IV over several hours to achieve the kind of concentrations required to be effective. (Merck has patented this formulation of the drug as Keytruda.)

“When you try to concentrate existing formulations, the molecules’ viscosity grows astronomically,” says Doyle. “At a critical point, they start almost feeling for each other, and interact to become a kind of paste.” Forced together, pembro molecules become unstable and change their structure, undermining their therapeutic properties.

So Doyle’s team of researchers in the Soft Matter Engineering Group set about creating a version of pembro that could be injected at high enough concentrations to be effective, but in small enough volumes to be administered comfortably and swiftly just under the skin (the second preference of most patients and clinicians, after swallowing a pill). With expertise in matters of flow, microfluidics, and pharmaceutical formulations, the lab was well-equipped for the challenge.

Go with the flow

“This MAB is super sticky and delicate, and we needed to find some way to get its molecules moving freely inside a syringe,” says Erfani. “The insight we had was to use hydrogel particles, made from sugar-based, water-loving biopolymers that provide a nice environment where a protein is going to be very happy,” says Doyle. “We’d used these for other applications, and I knew if we could make them small enough, they could get through a needle without clogging it.” 

The researchers knew from toxicity literature that their hydrogel capsules would be biocompatible, and would behave in a syringe. “The hydrogel particles are squeezy, and can roll over each other, and actually flow,” says Erfani. It looked like clear sailing to incorporate pembro molecules at the right density for a one- to two-millimeter subcutaneous injection. But, like so much in engineering, the devil turned out to be in the details.

“It was tricky keeping the antibody intact through the fabrication process, and then ensuring it could be biologically effective as it dispersed properly under the skin,” says Doyle. Any departure from the precise formulation of the pembro integrated into the soft hydrogel capsules might render the MAB ineffective, or worse.

In a series of experiments lasting nearly five years, Doyle’s lab experimented in achieving just the right balance of features. Their studies relied on a homemade device that jets out biopolymer solution and crystals of pembro together first into the air, and then into a bath where they fuse into beads.

“We tested many variables in our design space,” says Erfani, including different concentrations of pembro, and the composition and pH of the polymer solutions. “The goal wasn’t just developing a drug in our lab, but developing a process that could be easily adapted to pharmaceutical manufacturing.” With his prior industry background developing types of MAB in stable, crystalline structure, Erfani helped push the team over the finish line. “He not only brought all this physical chemistry to the process, but he figured out the experimental design and how to execute on it,” says Doyle.

A broad platform

The researchers are now putting their pembro formulation through its paces through in vivo trials, with the aim of U.S. Food and Drug Administration approval in the next few years. But Doyle and his group have broader goals for the hydrogel platform they invented. “We believe this platform is agnostic to the MAB, which means we can get a lot of different molecules formulated to the right concentration and flowability,” he says. “That’s a big deal.”

Among the possibilities Doyle envisions are a slow, sustained release of the MAB-containing hydrogel particles — think weeks — after injection. The platform could accommodate other kinds of molecules, such as cytokines, to amplify immune response, or target specific cancer pathways. Hydrogels could also incorporate two kinds of drugs that enhance each other’s properties.

Erfani points to the potential social impacts of the platform. “Our technology holds the possibility of improving the accessibility of treatments by reducing a patient’s dependency on hospitals,” he says. Replacing IV sessions with fewer, single-shot injections would free up time in clinics for more patients, encourage greater compliance, and even lower the price of the drug, he notes. People might someday administer their own injections at home.

Erfani is especially intrigued by the notion of moving many more drugs to this platform, including some that died early in development. “There are drugs companies that gave up because they couldn’t be formulated in high enough concentrations,” he says. “Wouldn’t it be super exciting to repurpose a lifesaving drug and bring it back to market?”

de MIT News https://ift.tt/B0hbyLS

Clothing brand helps give survivors of sexual violence a path forward

When Congolese doctor Denis Mukwege won a share of the Nobel Peace Prize in 2018, Milain Fayulu SM ’22 was filled with pride in his home country. He eagerly set an alarm from Miami to wake up in the early hours and watch Mukwege’s speech in Norway.

In the speech, Mukwege discussed his experience caring for tens of thousands of women who survived sexual violence during the civil wars in the Democratic Republic of Congo. Mukwege, who established the nonprofit Panzi Foundation to care for survivors and help them take back their lives, has called for an end to sexual violence in war.

“When you listen to that speech, you just shake,” Fayulu says. “I thought, ‘Is it really possible to develop a country that has this environment for women?’ It turns out there’s a clear correlation between how much freedom women have, how educated they are, and the development of the country.”

That insight was followed by another revelation about Mukwege’s message.

“The speech has relatively few views on YouTube,” Fayulu says. “It seemed like he was trapped in an echo chamber of folks passionate about international relations and politics. The reality is change doesn’t happen until your message starts permeating through the masses. I came to the conclusion that he wasn’t reaching the right audience.”

To help bring Mukwege’s message to the masses, Fayulu founded the Congo Clothing Company. The company sells Congo-inspired jackets, pants, T-shirts, and other apparel to a global customer base and donates a portion of proceeds to the Panzi Foundation’s job training efforts. The company also supports a course that teaches survivors of sexual violence how to sew.

“Nonprofits and nongovernmental organizations typically seek support through a damage-centered approach,” Fayulu explains. “We flip it by offering this cool brand with designs you might like. All we’re saying is ‘look good, do good.’ Get something you think is cool. The donation, the altruistic part, is baked into you buying something you already like. We’re inviting you to join us from a place of culture and fashion.”

Congo Clothing Company’s merchandise is shipped with a booklet that tells Congo’s history grappling with war, educating customers in the hopes they become ambassadors for the company’s mission.

“When you get the jacket or shirt, you enter the ecosystem,” Fayulu says. “When you wear it around, say, Cambridge, people will stop you — we hear about it all the time — and you’ll have no option but to tell the story of the women in Congo and how your purchase has made an impact. That way the story — and the support — scales. We don’t need The New York Times to report on it, we just need Gen Zers and Millennials around the world to take ownership of the problem and do something to help these women get back on track.”

Building a brand at MIT

Although Fayulu was born in the Democratic Republic of Congo, he lived throughout Africa in his youth before finishing high school in Paris. He attended the University of Miami as an undergraduate, where he began his first forays into entrepreneurship, founding an online skincare brand for people of color and then a property-tech company.

After hearing Mukwege’s moving speech, Fayulu worked for a year to meet with the doctor, eventually tracking him down in Los Angeles in 2019.

He told Mukwege about the shift to conscious consumerism in Western markets and explained his idea to help take Mukwege’s message to a wider audience.

“People want brands that align with their values,” Fayulu says. “Congo Clothing Company was really born from this idea that if we can find a way to merge his work with the mass market, people with a voice in their community, people with followers on social media,  you could democratize his work and get it to scale.”

Fayulu came to MIT in 2020 to pursue a lifelong interest in political science while also leveraging the Institute’s entrepreneurship resources to get CCC off the ground.

First, he received a fellowship from the Legatum Center at MIT, which support social entrepreneurship. Then he took courses in D-Lab, received guidance from MIT’s Venture Mentoring Service, and frequented the Martin Trust Center for MIT Entrepreneurship, through which he entered entrepreneurship programs including the MIT Fuse bootcamp, StartMIT, and the delta v summer accelerator.

Through it all, Fayulu refined his idea for Congo Clothing Company and advanced the brand.

“Education is important,” he says. “We think of ourselves as storytellers. We’re competing for attention. We’ve chosen fashion as our medium, and our ability to amplify the message rests in our ability to give people a compelling story that they can then internalize, own, and regurgitate. That’s how we spread the message.”

Congo Clothing Company’s logo is a zigzag in between two straight lines. Fayulu says it represents the merging of an ancient Congolese Kingdom known as Kuba (the zigzag) with a Western aesthetic (the straight lines). As the brand expands its product line, Fayulu has big dreams for the company.

“If you walk down the street in Boston or Cambridge, you can count the number of Patagonia jackets you see,” Fayulu says. “We want people to see that zigzag and for it to become a recognizable logo. Africa doesn’t have many ubiquitous brands, but that’s what we want Congo Clothing Company to become. We want to become a part of people’s daily lives.”

Empowering survivors

The Panzi Hospital opened in 1999. It has since expanded to provide not only medical care but also psychological treatment, legal support, and job training. To date, it has helped more than 85,000 women.

Congo Clothing Company worked with Panzi to develop part of the job training curriculum to teach women how to sew. To date the company has funded more than 7,000 training days.

“If you don’t have anything when you get out of the hospital, you’ll never recover,” Fayulu says. “Once you complete the workshop, you have a skill and you can go back into society with a toolkit. Now survivors have a way to make a living, and survivors with kids can put them through school so that they’re not subject to being recruited by rebel groups. That way the violence decreases and you create a virtuous cycle in the communities.”

Eventually CCC wants to expand the partnership to have women manufacture and design parts of its clothing. In the meantime, with the help of Enrique Avina ’22 and current MIT undergraduate Rithvik Ganesh, CCC aims to deliver more impact digitally.

In that spirit, the company is developing an app it’s calling the ‘CCC Creator Studio’ to provide a faster, more direct stream of income for the women.

“Shipping garments is a heavy lift,” Fayulu says. “Shipping designs is easier.”

All of CCC’s initiatives are distinct from the aid-based support vulnerable communities typically receive.

“There’s this savior complex, the idea is we’re going to shower these people with money and food, and we’ll feel good about ourselves, and that’s it,” Fayulu says. “Everyone welcomes aid if they need it, but what really changes things is when you teach people new skills. When you talk to these women, they want something they can run independently so they can rely on themselves. These women are not helpless — these women are very resilient. They’re some of the most resilient people out there.”

de MIT News https://ift.tt/hJCdOQi

A method for designing neural networks optimally suited for certain tasks

Neural networks, a type of machine-learning model, are being used to help humans complete a wide variety of tasks, from predicting if someone’s credit score is high enough to qualify for a loan to diagnosing whether a patient has a certain disease. But researchers still have only a limited understanding of how these models work. Whether a given model is optimal for certain task remains an open question.

MIT researchers have found some answers. They conducted an analysis of neural networks and proved that they can be designed so they are “optimal,” meaning they minimize the probability of misclassifying borrowers or patients into the wrong category when the networks are given a lot of labeled training data. To achieve optimality, these networks must be built with a specific architecture.

The researchers discovered that, in certain situations, the building blocks that enable a neural network to be optimal are not the ones developers use in practice. These optimal building blocks, derived through the new analysis, are unconventional and haven’t been considered before, the researchers say.

In a paper published this week in the Proceedings of the National Academy of Sciences, they describe these optimal building blocks, called activation functions, and show how they can be used to design neural networks that achieve better performance on any dataset. The results hold even as the neural networks grow very large. This work could help developers select the correct activation function, enabling them to build neural networks that classify data more accurately in a wide range of application areas, explains senior author Caroline Uhler, a professor in the Department of Electrical Engineering and Computer Science (EECS).

“While these are new activation functions that have never been used before, they are simple functions that someone could actually implement for a particular problem. This work really shows the importance of having theoretical proofs. If you go after a principled understanding of these models, that can actually lead you to new activation functions that you would otherwise never have thought of,” says Uhler, who is also co-director of the Eric and Wendy Schmidt Center at the Broad Institute of MIT and Harvard, and a researcher at MIT’s Laboratory for Information and Decision Systems (LIDS) and its Institute for Data, Systems and Society (IDSS).

Joining Uhler on the paper are lead author Adityanarayanan Radhakrishnan, an EECS graduate student and an Eric and Wendy Schmidt Center Fellow, and Mikhail Belkin, a professor in the Halicioğlu Data Science Institute at the University of California at San Diego.

Activation investigation

A neural network is a type of machine-learning model that is loosely based on the human brain. Many layers of interconnected nodes, or neurons, process data. Researchers train a network to complete a task by showing it millions of examples from a dataset.

For instance, a network that has been trained to classify images into categories, say dogs and cats, is given an image that has been encoded as numbers. The network performs a series of complex multiplication operations, layer by layer, until the result is just one number. If that number is positive, the network classifies the image a dog, and if it is negative, a cat.

Activation functions help the network learn complex patterns in the input data. They do this by applying a transformation to the output of one layer before data are sent to the next layer. When researchers build a neural network, they select one activation function to use. They also choose the width of the network (how many neurons are in each layer) and the depth (how many layers are in the network.)

“It turns out that, if you take the standard activation functions that people use in practice, and keep increasing the depth of the network, it gives you really terrible performance. We show that if you design with different activation functions, as you get more data, your network will get better and better,” says Radhakrishnan.

He and his collaborators studied a situation in which a neural network is infinitely deep and wide — which means the network is built by continually adding more layers and more nodes — and is trained to perform classification tasks. In classification, the network learns to place data inputs into separate categories.

“A clean picture”

After conducting a detailed analysis, the researchers determined that there are only three ways this kind of network can learn to classify inputs. One method classifies an input based on the majority of inputs in the training data; if there are more dogs than cats, it will decide every new input is a dog. Another method classifies by choosing the label (dog or cat) of the training data point that most resembles the new input.

The third method classifies a new input based on a weighted average of all the training data points that are similar to it. Their analysis shows that this is the only method of the three that leads to optimal performance. They identified a set of activation functions that always use this optimal classification method.

“That was one of the most surprising things — no matter what you choose for an activation function, it is just going to be one of these three classifiers. We have formulas that will tell you explicitly which of these three it is going to be. It is a very clean picture,” he says.

They tested this theory on a several classification benchmarking tasks and found that it led to improved performance in many cases. Neural network builders could use their formulas to select an activation function that yields improved classification performance, Radhakrishnan says.

In the future, the researchers want to use what they’ve learned to analyze situations where they have a limited amount of data and for networks that are not infinitely wide or deep. They also want to apply this analysis to situations where data do not have labels.

“In deep learning, we want to build theoretically grounded models so we can reliably deploy them in some mission-critical setting. This is a promising approach at getting toward something like that — building architectures in a theoretically grounded way that translates into better results in practice,” he says.

This work was supported, in part, by the National Science Foundation, Office of Naval Research, the MIT-IBM Watson AI Lab, the Eric and Wendy Schmidt Center at the Broad Institute, and a Simons Investigator Award.

de MIT News https://ift.tt/T7RvDGb

New nanoparticles can perform gene editing in the lungs

Engineers at MIT and the University of Massachusetts Medical School have designed a new type of nanoparticle that can be administered to the lungs, where it can deliver messenger RNA encoding useful proteins.

With further development, these particles could offer an inhalable treatment for cystic fibrosis and other diseases of the lung, the researchers say.

“This is the first demonstration of highly efficient delivery of RNA to the lungs in mice. We are hopeful that it can be used to treat or repair a range of genetic diseases, including cystic fibrosis,” says Daniel Anderson, a professor in MIT’s Department of Chemical Engineering and a member of MIT’s Koch Institute for Integrative Cancer Research and Institute for Medical Engineering and Science (IMES).

In a study of mice, Anderson and his colleagues used the particles to deliver mRNA encoding the machinery needed for CRISPR/Cas9 gene editing. That could open the door to designing therapeutic nanoparticles that can snip out and replace disease-causing genes.

The senior authors of the study, which appears today in Nature Biotechnology, are Anderson; Robert Langer, the David H. Koch Institute Professor at MIT; and Wen Xue, an associate professor at the UMass Medical School RNA Therapeutics Institute. Bowen Li, a former MIT postdoc who is now an assistant professor at the University of Toronto; Rajith Singh Manan, an MIT postdoc; and Shun-Qing Liang, a postdoc at UMass Medical School, are paper’s lead authors.

Targeting the lungs

Messenger RNA holds great potential as a therapeutic for treating a variety of diseases caused by faulty genes. One obstacle to its deployment thus far has been difficulty in delivering it to the right part of the body, without off-target effects. Injected nanoparticles often accumulate in the liver, so several clinical trials evaluating potential mRNA treatments for diseases of the liver are now underway. RNA-based Covid-19 vaccines, which are injected directly into muscle tissue, have also proven effective. In many of those cases, mRNA is encapsulated in a lipid nanoparticle — a fatty sphere that protects mRNA from being broken down prematurely and helps it enter target cells.

Several years ago, Anderson’s lab set out to design particles that would be better able to transfect the epithelial cells that make up most of the lining of the lungs. In 2019, his lab created nanoparticles that could deliver mRNA encoding a bioluminescent protein to lung cells. Those particles were made from polymers instead of lipids, which made them easier to aerosolize for inhalation into the lungs. However, more work is needed on those particles to increase their potency and maximize their usefulness.

In their new study, the researchers set out to develop lipid nanoparticles that could target the lungs. The particles are made up of molecules that contain two parts: a positively charged headgroup and a long lipid tail. The positive charge of the headgroup helps the particles to interact with negatively charged mRNA, and it also help mRNA to escape from the cellular structures that engulf the particles once they enter cells.

The lipid tail structure, meanwhile, helps the particles to pass through the cell membrane. The researchers came up with 10 different chemical structures for the lipid tails, along with 72 different headgroups. By screening different combinations of these structures in mice, the researchers were able to identify those that were most likely to reach the lungs.

Efficient delivery

In further tests in mice, the researchers showed that they could use the particles to deliver mRNA encoding CRISPR/Cas9 components designed to cut out a stop signal that was genetically encoded into the animals’ lung cells. When that stop signal is removed, a gene for a fluorescent protein turns on. Measuring this fluorescent signal allows the researchers to determine what percentage of the cells successfully expressed the mRNA.

After one dose of mRNA, about 40 percent of lung epithelial cells were transfected, the researchers found. Two doses brought the level to more than 50 percent, and three doses up to 60 percent. The most important targets for treating lung disease are two types of epithelial cells called club cells and ciliated cells, and each of these was transfected at about 15 percent.

“This means that the cells we were able to edit are really the cells of interest for lung disease,” Li says. “This lipid can enable us to deliver mRNA to the lung much more efficiently than any other delivery system that has been reported so far.”

The new particles also break down quickly, allowing them to be cleared from the lung within a few days and reducing the risk of inflammation. The particles could also be delivered multiple times to the same patient if repeat doses are needed. This gives them an advantage over another approach to delivering mRNA, which uses a modified version of harmless adenoviruses. Those viruses are very effective at delivering RNA but can’t be given repeatedly because they induce an immune response in the host.

“This achievement paves the way for promising therapeutic lung gene delivery applications for various lung diseases,” says Dan Peer, director of the Laboratory of Precision NanoMedicine at Tel Aviv University, who was not involved in the study. “This platform holds several advantages compared to conventional vaccines and therapies, including that it’s cell-free, enables rapid manufacturing, and has high versatility and a favorable safety profile.”

To deliver the particles in this study, the researchers used a method called intratracheal instillation, which is often used as a way to model delivery of medication to the lungs. They are now working on making their nanoparticles more stable, so they could be aerosolized and inhaled using a nebulizer.

The researchers also plan to test the particles to deliver mRNA that could correct the genetic mutation found in the gene that causes cystic fibrosis, in a mouse model of the disease. They also hope to develop treatments for other lung diseases, such as idiopathic pulmonary fibrosis, as well as mRNA vaccines that could be delivered directly to the lungs.

The research was funded by Translate Bio, the National Institutes of Health, the Leslie Dan Faculty of Pharmacy startup fund, a PRiME Postdoctoral Fellowship from the University of Toronto, the American Cancer Society, and the Cystic Fibrosis Foundation.

de MIT News https://ift.tt/ID7FskJ

Learning to design with atoms and molecules

MIT undergraduates are learning about nanoscale science and engineering from individual atoms up to full-scale functional systems, and they’re doing it hands-on at MIT.nano.

In class 6.2540 (Nanotechnology: From Atoms to Systems) students spend over nine weeks inside MIT.nano’s labs, learning basic skills that allow them to apply their knowledge of the nanoscale to design and build spectrometers, make quantum dots, fabricate light-emitting diodes (LEDs) and tunneling chemical sensors, and test and package their sensors into active displays and systems.

Bringing the science to life in this way has generated much excitement among the undergraduates. Dahlia Dry, a senior majoring in physics, said her faculty advisor suggested the class would show her the fun in quantum mechanics. “He was right. This class was exactly what middle-school-aged me thought MIT would be like, in all the best ways,” she says.

Word must be getting out about the fun as the class is drawing interest from undergraduates majoring in many different subjects. In fall 2022, six academic departments were represented by the 23 students enrolled.

“This class is quintessentially an ‘MIT’ class,” says Neil Deshmukh, an EECS junior. “Since coming to campus, I've always wanted to take a class where we were free to build nearly any idea, with access to state-of-the-art equipment and amazing instructors. In 6.2540, that's exactly what we did, and it was one of the best experiences I've had.”

The class is taught by three EECS professors: Farnaz Niroui, the EE Landsman Career Development Assistant Professor; Rajeev Ram, professor of electrical engineering; and Tayo Akinwande, the Thomas and Gerd Perkins Professor of Electrical Engineering and Computer Science.

“In this class we take a design approach, rather than the more common abstract and theoretical style,” explains Niroui. “We teach the fundamentals of quantum mechanics and nanoscale science by directly relating them to the design and engineering of diverse technologies.”

For this reason, the lectures are closely integrated with design projects and weekly lab modules. Starting the very first week, the students are inside the lab, learning to work in a cleanroom and acquiring the basic nanofabrication, processing, and characterization skills to investigate and implement concepts they have learned in the lectures — from fundamental science to material synthesis, device design, and full systems integration.

Rather than watching staff run the equipment, the undergraduates do the work themselves using simplified engineering and fabrication flows. “This was the most fascinating class I have taken at MIT, and that's despite it being in an area that I knew nothing about beforehand,” says EECS sophomore Eric Zhang. “It opened my eyes to an entire research and engineering field that I would never have known about otherwise.”

Each week’s lab work builds off the ones before, starting at the nano- and micro-level and building up to full-scale devices. Students learn about light-matter interactions and build their own microscopes and spectrometers, then use their new tools to characterize the materials and devices they make throughout the term. Further into the semester, they investigate the power of quantum mechanics and the design of nanomaterials through chemical synthesis of quantum dots, tuning their emission color by controlling their size. The following week, they use quantum dots to design and make an LED. This lab is followed by design and fabrication of a quantum tunneling chemical sensor based on graphene-polymer composites. In the final lab, the students use these LEDs and tunneling sensors to integrate a pixelated LED display into a handheld sensor-display system.

For their end-of-semester projects, the students split into teams to design and build something entirely from scratch, provided their idea uses the science, materials, and techniques covered in the class and has at least one feature smaller than 100 nanometers. In the fall 2022 semester, the undergraduates fabricated memristors for next-generation unconventional computing; nature-inspired structured lenses to improve LED efficiency; flexible graphene supercapacitors for solar energy storage; a flexible pulse oximeter; tandem solar cells based on band-gap engineering; and a transistor using atomically-thin 2D materials.

In addition to hands-on experience using tools for nanoscale engineering inside MIT.nano’s cleanroom and other labs, 6.2540 provides the opportunity for undergraduates to present at the Microsystems Annual Research Conference (MARC), co-sponsored by the Microsystems Technology Laboratories and MIT.nano. The long-standing event, which brings together over 200 MIT faculty, students, and industry partners each year, traditionally features graduate-level research.

de MIT News https://ift.tt/tEwiSdq

Boosting passenger experience and increasing connectivity at the Hong Kong International Airport

Recently, a cohort of 36 students from MIT and universities across Hong Kong came together for the MIT Entrepreneurship and Maker Skills Integrator (MEMSI), an intense two-week startup boot camp hosted at the MIT Hong Kong Innovation Node.

“We’re very excited to be in Hong Kong,” said Professor Charles Sodini, LeBel Professor of Electrical Engineering and faculty director of the Node. “The dream always was to bring MIT and Hong Kong students together.”

Students collaborated on six teams to meet real-world industry challenges through action learning, defining a problem, designing a solution, and crafting a business plan. The experience culminated in the MEMSI Showcase, where each team presented its process and unique solution to a panel of judges. “The MEMSI program is a great demonstration of important international educational goals for MIT,” says Professor Richard Lester, associate provost for international activities and chair of the Node Steering Committee at MIT. “It creates opportunities for our students to solve problems in a particular and distinctive cultural context, and to learn how innovations can cross international boundaries.” 

Meeting an urgent challenge in the travel and tourism industry

The Hong Kong Airport Authority (AAHK) served as the program’s industry partner for the third consecutive year, challenging students to conceive innovative ideas to make passenger travel more personalized from end-to-end while increasing connectivity. As the travel industry resuscitates profitability and welcomes crowds back amidst ongoing delays and labor shortages, the need for a more passenger-centric travel ecosystem is urgent.

The airport is the third-busiest international passenger airport and the world’s busiest cargo transit. Students experienced an insider’s tour of the Hong Kong International Airport to gain on-the-ground orientation. They observed firsthand the complex logistics, possibilities, and constraints of operating with a team of 78,000 employees who serve 71.5 million passengers with unique needs and itineraries.

Throughout the program, the cohort was coached and supported by MEMSI alumni, travel industry mentors, and MIT faculty such as Richard de Neufville, professor of engineering systems.

The mood inside the open-plan MIT Hong Kong Innovation Node was nonstop energetic excitement for the entire program. Each of the six teams was composed of students from MIT and from Hong Kong universities. They learned to work together under time pressure, develop solutions, receive feedback from industry mentors, and iterate around the clock.

“MEMSI was an enriching and amazing opportunity to learn about entrepreneurship while collaborating with a diverse team to solve a complex problem,” says Maria Li, a junior majoring in computer science, economics, and data science at MIT. “It was incredible to see the ideas we initially came up with as a team turn into a single, thought-out solution by the end.”

Unsurprisingly given MIT’s focus on piloting the latest technology and the tech-savvy culture of Hong Kong as a global center, many team projects focused on virtual reality, apps, and wearable technology designed to make passengers’ journeys more individualized, efficient, or enjoyable.

After observing geospatial patterns charting passengers’ movement through an airport, one team realized that many people on long trips aim to meet fitness goals by consciously getting their daily steps power walking the expansive terminals. The team’s prototype, FitAir, is a smart, biometric token integrated virtual coach, which plans walking routes within the airport to promote passenger health and wellness.

Another team noted a common frustration among frequent travelers who manage multiple mileage rewards program profiles, passwords, and status reports. They proposed AirPoint, a digital wallet that consolidates different rewards programs and presents passengers with all their airport redemption opportunities in one place.

“Today, there is no loser,” said Vivian Cheung, chief operating officer of AAHK, who served as one of the judges. “Everyone is a winner. I am a winner, too. I have learned a lot from the showcase. Some of the ideas, I believe, can really become a business.”

Cheung noted that in just 12 days, all teams observed and solved her organization’s pain points and successfully designed solutions to address them.

More than a competition

Although many of the models pitched are inventive enough to potentially shape the future of travel, the main focus of MEMSI isn’t to act as yet another startup challenge and incubator.

“What we’re really focusing on is giving students the ability to learn entrepreneurial thinking,” explains Marina Chan, senior director and head of education at the Node. “It’s the dynamic experience in a highly connected environment that makes being in Hong Kong truly unique. When students can adapt and apply theory to an international context, it builds deeper cultural competency.”

From an aerial view, the boot camp produced many entrepreneurs in the making and lasting friendships, and respect for other cultural backgrounds and operating environments.

“I learned the overarching process of how to make a startup pitch, all the way from idea generation, market research, and making business models, to the pitch itself and the presentation,” says Arun Wongprommoon, a senior double majoring in computer science and engineering and linguistics.  “It was all a black box to me before I came into the program.”

He said he gained tremendous respect for the startup world and the pure hard work and collaboration required to get ahead.

Spearheaded by the Node, MEMSI is a collaboration among the MIT Innovation Initiative, the Martin Trust Center for Entrepreneurship, the MIT International Science and Technology Initiatives, and Project Manus. Learn more about applying to MEMSI.

de MIT News https://ift.tt/LknlNiI

Bacterial injection system delivers proteins in mice and human cells

Researchers at the McGovern Institute for Brain Research at MIT and the Broad Institute of MIT and Harvard have harnessed a natural bacterial system to develop a new protein delivery approach that works in human cells and animals. The technology, described today in Nature, can be programmed to deliver a variety of proteins, including ones for gene editing, to different cell types. The system could potentially be a safe and efficient way to deliver gene therapies and cancer therapies.

Led by MIT Associate Professor Feng Zhang, who is a McGovern Institute investigator and Broad Institute core member, the team took advantage of a tiny syringe-like injection structure, produced by a bacterium, that naturally binds to insect cells and injects a protein payload into them. The researchers used the artificial intelligence tool AlphaFold to engineer these syringe structures to deliver a range of useful proteins to both human cells and cells in live mice.

“This is a really beautiful example of how protein engineering can alter the biological activity of a natural system,” says Joseph Kreitz, the study’s first author, a graduate student in biological engineering at MIT, and a member of Zhang’s lab. “I think it substantiates protein engineering as a useful tool in bioengineering and the development of new therapeutic systems.”

“Delivery of therapeutic molecules is a major bottleneck for medicine, and we will need a deep bench of options to get these powerful new therapies into the right cells in the body,” adds Zhang. “By learning from how nature transports proteins, we were able to develop a new platform that can help address this gap.”

Zhang is senior author on the study and is also the James and Patricia Poitras Professor of Neuroscience at MIT and an investigator at the Howard Hughes Medical Institute.

Injection via contraction

Symbiotic bacteria use the roughly 100-nanometer-long syringe-like machines to inject proteins into host cells to help adjust the biology of their surroundings and enhance their survival. These machines, called extracellular contractile injection systems (eCISs), consist of a rigid tube inside a sheath that contracts, driving a spike on the end of the tube through the cell membrane. This forces protein cargo inside the tube to enter the cell.

On the outside of one end of the eCIS are tail fibers that recognize specific receptors on the cell surface and latch on. Previous research has shown that eCISs can naturally target insect and mouse cells, but Kreitz thought it might be possible to modify them to deliver proteins to human cells by re-engineering the tail fibers to bind to different receptors.

Using AlphaFold, which predicts a protein’s structure from its amino acid sequence, the researchers redesigned tail fibers of an eCIS produced by Photorhabdus bacteria to bind to human cells. By re-engineering another part of the complex, the scientists tricked the syringe into delivering a protein of their choosing, in some cases with remarkably high efficiency.

The team made eCISs that targeted cancer cells expressing the EGF receptor and showed that they killed almost 100 percent of the cells, but did not affect cells without the receptor. Though efficiency depends in part on the receptor the system is designed to target, Kreitz says that the findings demonstrate the promise of the system with thoughtful engineering.

The researchers also used an eCIS to deliver proteins to the brain in live mice — where it didn’t provoke a detectable immune response, suggesting that eCISs could one day be used to safely deliver gene therapies to humans.

Packaging proteins

Kreitz says the eCIS system is versatile, and the team has already used it to deliver a range of cargoes including base editor proteins (which can make single-letter changes to DNA), proteins that are toxic to cancer cells, and Cas9, a large DNA-cutting enzyme used in many gene editing systems.

In the future, Kreitz says researchers could engineer other components of the eCIS system to tune other properties, or to deliver other cargoes such as DNA or RNA. He also wants to better understand the function of these systems in nature.

“We and others have shown that this type of system is incredibly diverse across the biosphere, but they are not very well characterized,” Kreitz said. “And we believe this type of system plays really important roles in biology that are yet to be explored.”

This work was supported, in part, by the National Institutes of Health, Howard Hughes Medical Institute, Poitras Center for Psychiatric Disorders Research at MIT, Hock E. Tan and K. Lisa Yang Center for Autism Research at MIT, K. Lisa Yang and Hock E. Tan Molecular Therapeutics Center at MIT, K. Lisa Yang Brain-Body Center at MIT, Broad Institute Programmable Therapeutics Gift Donors, The Pershing Square Foundation, William Ackman, Neri Oxman, J. and P. Poitras, Kenneth C. Griffin, BT Charitable Foundation, the Asness Family Foundation, the Phillips family, D. Cheng, and R. Metcalfe.

de MIT News https://ift.tt/3wzOprM

New algorithm keeps drones from colliding in midair

When multiple drones are working together in the same airspace, perhaps spraying pesticide over a field of corn, there’s a risk they might crash into each other.

To help avoid these costly crashes, MIT researchers presented a system called MADER in 2020. This multiagent trajectory-planner enables a group of drones to formulate optimal, collision-free trajectories. Each agent broadcasts its trajectory so fellow drones know where it is planning to go. Agents then consider each other’s trajectories when optimizing their own to ensure they don’t collide.

But when the team tested the system on real drones, they found that if a drone doesn’t have up-to-date information on the trajectories of its partners, it might inadvertently select a path that results in a collision. The researchers revamped their system and are now rolling out Robust MADER, a multiagent trajectory planner that generates collision-free trajectories even when communications between agents are delayed.

“MADER worked great in simulations, but it hadn’t been tested in hardware. So, we built a bunch of drones and started flying them. The drones need to talk to each other to share trajectories, but once you start flying, you realize pretty quickly that there are always communication delays that introduce some failures,” says Kota Kondo, an aeronautics and astronautics graduate student.

The algorithm incorporates a delay-check step during which a drone waits a specific amount of time before it commits to a new, optimized trajectory. If it receives additional trajectory information from fellow drones during the delay period, it might abandon its new trajectory and start the optimization process over again.

When Kondo and his collaborators tested Robust MADER, both in simulations and flight experiments with real drones, it achieved a 100 percent success rate at generating collision-free trajectories. While the drones’ travel time was a bit slower than it would be with some other approaches, no other baselines could guarantee safety.

“If you want to fly safer, you have to be careful, so it is reasonable that if you don’t want to collide with an obstacle, it will take you more time to get to your destination. If you collide with something, no matter how fast you go, it doesn’t really matter because you won’t reach your destination,” Kondo says.  

Kondo wrote the paper with Jesus Tordesillas, a postdoc; Parker C. Lusk, a graduate student; Reinaldo Figueroa, Juan Rached, and Joseph Merkel, MIT undergraduates; and senior author Jonathan P. How, the Richard C. Maclaurin Professor of Aeronautics and Astronautics and a member of the MIT-IBM Watson AI Lab. The research will be presented at the International Conference on Robots and Automation.

Planning trajectories

MADER is an asynchronous, decentralized, multiagent trajectory-planner. This means that each drone formulates its own trajectory and that, while all agents must agree on each new trajectory, they don’t need to agree at the same time. This makes MADER more scalable than other approaches, since it would be very difficult for thousands of drones to agree on a trajectory simultaneously. Due to its decentralized nature, the system would also work better in real-world environments where drones may fly far from a central computer.

With MADER, each drone optimizes a new trajectory using an algorithm that incorporates the trajectories it has received from other agents. By continually optimizing and broadcasting their new trajectories, the drones avoid collisions.

But perhaps one agent shared its new trajectory several seconds ago, but a fellow agent didn’t receive it right away because the communication was delayed. In real-world environments, signals are often delayed by interference from other devices or environmental factors like stormy weather. Due to this unavoidable delay, a drone might inadvertently commit to a new trajectory that sets it on a collision course.

Robust MADER prevents such collisions because each agent has two trajectories available. It keeps one trajectory that it knows is safe, which it has already checked for potential collisions. While following that original trajectory, the drone optimizes a new trajectory but does not commit to the new trajectory until it completes a delay-check step.

During the delay-check period, the drone spends a fixed amount of time repeatedly checking for communications from other agents to see if its new trajectory is safe. If it detects a potential collision, it abandons the new trajectory and starts the optimization process over again.

The length of the delay-check period depends on the distance between agents and environmental factors that could hamper communications, Kondo says. If the agents are many miles apart, for instance, then the delay-check period would need to be longer.

Completely collision-free

The researchers tested their new approach by running hundreds of simulations in which they artificially introduced communication delays. In each simulation, Robust MADER was 100 percent successful at generating collision-free trajectories, while all the baselines caused crashes.

The researchers also built six drones and two aerial obstacles and tested Robust MADER in a multiagent flight environment. They found that, while using the original version of MADER in this environment would have resulted in seven collisions, Robust MADER did not cause a single crash in any of the hardware experiments.

“Until you actually fly the hardware, you don’t know what might cause a problem. Because we know that there is a difference between simulations and hardware, we made the algorithm robust, so it worked in the actual drones, and seeing that in practice was very rewarding,” Kondo says.

Drones were able to fly 3.4 meters per second with Robust MADER, although they had a slightly longer average travel time than some baselines. But no other method was perfectly collision-free in every experiment.

In the future, Kondo and his collaborators want to put Robust MADER to the test outdoors, where many obstacles and types of noise can affect communications. They also want to outfit drones with visual sensors so they can detect other agents or obstacles, predict their movements, and include that information in trajectory optimizations.

This work was supported by Boeing Research and Technology.

de MIT News https://ift.tt/vUSXtpW

Fieldwork class examines signs of climate change in Hawaii

When Joy Domingo-Kameenui spent two weeks in her native Hawaii as part of MIT class 1.091 (Traveling Research Environmental eXperiences), she was surprised to learn about the number of invasive and endangered species. “I knew about Hawaiian ecology from middle and high school but wasn’t fully aware to the extent of how invasive species and diseases have resulted in many of Hawaii’s endemic species becoming threatened,” says Domingo-Kameenui.  

Domingo-Kameenui was part of a group of MIT students who conducted field research on the Big Island of Hawaii in the Traveling Research Environmental eXperiences (TREX) class offered by the Department of Civil and Environmental Engineering. The class provides undergraduates an opportunity to gain hands-on environmental fieldwork experience using Hawaii’s geology, chemistry, and biology to address two main topics of climate change concern: sulfur dioxide (SO₂) emissions and forest health.

“Hawaii is this great system for studying the effects of climate change,” says David Des Marais, the Cecil and Ida Green Career Development Professor of Civil and Environmental Engineering and lead instructor of TREX. “Historically, Hawaii has had occasional mild droughts that are related to El Niño, but the droughts are getting stronger and more frequent. And we know these types of extreme weather events are going to happen worldwide.”

Climate change impacts on forests

The frequency and intensity of extreme events are also becoming more of a problem for forests and plant life. Forests have a certain distribution of vegetation and as you get higher in elevation, the trees gradually turn into shrubs, and then rock. Trees don’t grow above the timberline, where the temperature and precipitation changes dramatically at the high elevations. “But unlike the Sierra Nevada or the Rockies, where the trees gradually change as you go up the mountains, in Hawaii, they gradually change, and then they just stop,” says Des Marais.

“Why this is an interesting model for climate change,” explains Des Marais, “is that line where trees stop [growing] is going to move, and it’s going to become more unstable as the trade winds are affected by global patterns of air circulation, which are changing because of climate change.”

The research question that Des Marais asks students to explore — How is the Hawaiian forest going to be affected by climate change? — uses Hawaii as a model for broader patterns in climate change for forests.

To dive deeper into this question, students trekked up the mountain taking ground-level measurements of canopy cover with a camera app on their cellphones, estimating how much tree coverage blankets the sky when looking up, and observing how the canopy cover thins until they see no tree coverage at all as they go further up the mountain. Drones also flew above the forest to measure chlorophyll and how much plant matter remains. And then satellite data products from NASA and the European Space Agency were used to measure the distribution of chlorophyll, climate, and precipitation data from space.

They also worked directly with community stakeholders at three locations around the island to access the forests and use technology to assess the ecology and biodiversity challenges. One of those stakeholders was the Kamehameha Schools Natural and Cultural Ecosystems Division, whose mission is to preserve the land and manage it in a sustainable way. Students worked with their plant biologists to help address and think about what management decisions will support the future health of their forests.

“Across the island, rising temperatures and abnormal precipitation patterns are the main drivers of drought, which really has significant impacts on biodiversity, and overall human health,” says Ava Gillikin, a senior in civil and environmental engineering.

Gillikin adds that “a good proportion of the island’s water system relies on rainwater catchment, exposing vulnerabilities to fluctuations in rain patterns that impact many people’s lives.”

Deadly threats to native plants

The other threats to Hawaii’s forests are invasive species causing ecological harm, from the prevalence of non-indigenous mosquitoes leading to increases in avian malaria and native bird death that threaten the native ecosystem, to a plant called strawberry guava.

Strawberry guava is taking over Hawaii’s native ōhiʻa trees, which Domingo-Kameenui says is also contributing to Hawaii’s water production. “The plants absorb water quickly so there’s less water runoff for groundwater systems.”

A fungal pathogen is also infecting native ōhiʻa trees. The disease, called rapid ʻohiʻa death (ROD), kills the tree within a few days to weeks. The pathogen was identified by researchers on the island in 2014 from the fungal genus, Ceratocystis. The fungal pathogen was likely carried into the forests by humans on their shoes, or contaminated tools, gear, and vehicles traveling from one location to another. The fungal disease is also transmitted by beetles that bore into trees and create a fine powder-like dust. This dust from an infected tree is then mixed with the fungal spores and can easily spread to other trees by wind, or contaminated soil.

For Gillikin, seeing the effects of ROD in the field highlighted the impact improper care and preparation can have on native forests. “The ‘ohi’a tree is one of the most prominent native trees, and ROD can kill the trees very rapidly by putting a strain on its vascular system and preventing water from reaching all parts of the tree,” says Gillikin.

Before entering the forests, students sprayed their shoes and gear with ethanol frequently to prevent the spread.

Uncovering chemical and particle formation

A second research project in TREX studied volcanic smog (vog) that plagues the air, making visibility problematic at times and causing a lot of health problems for people in Hawaii. The active Kilauea volcano releases SO₂ into the atmosphere. When the SO₂ mixes with other gasses emitted from the volcano and interacts with sunlight and the atmosphere, particulate matter forms.

Students in the Kroll Group, led by Jesse Kroll, professor of civil and environmental engineering and chemical engineering, have been studying SO₂ and particulate matter over the years, but not the chemistry directly in how those chemical transformations occur.

“There's a hypothesis that there is a functional connection between the SO₂ and particular matter, but that's never been directly demonstrated,” says Des Marais.

Testing that hypothesis, the students were able to measure two different sizes of particulate matter formed from the SO₂ and develop a model to show how much vog is generated downstream of the volcano.

They spent five days at two sites from sunrise to late morning measuring particulate matter formation as the sun comes up and starts creating new particles. Using a combination of data sources for meteorology, such as UV index, wind speed, and humidity, the students built a model that demonstrates all the pieces of an equation that can calculate when new particles are formed.

“You can build what you think that equation is based on first-principle understanding of the chemical composition, but what they did was measured it in real time with measurements of the chemical reagents,” says Des Marias.

The students measured what was going to catalyze the chemical reaction of particulate matter — for instance, things like sunlight and ozone — and then calculated numbers to the outputs.

“What they found, and what seems to be happening, is that the chemical reagents are accumulating overnight,” says Des Marais. “Then as soon as the sun rises in the morning all the transformation happens in the atmosphere. A lot of the reagents are used up and the wind blows everything away, leaving the other side of the island with polluted air,” adds Des Marais.

“I found the vog particle formation fieldwork a surprising research learning,” adds Domingo-Kameenui who did some atmospheric chemistry research in the Kroll Group. “I just thought particle formation happened in the air, but we found wind direction and wind speed at a certain time of the day was extremely important to particle formation. It’s not just chemistry you need to look at, but meteorology and sunlight,” she adds.

Both Domingo-Kameenui and Gillikin found the fieldwork class an important and memorable experience with new insight that they will carry with them beyond MIT.  

How Gillikin approaches fieldwork or any type of community engagement in another culture is what she will remember most. “When entering another country or culture, you are getting the privilege to be on their land, to learn about their history and experiences, and to connect with so many brilliant people,” says Gillikin. “Everyone we met in Hawaii had so much passion for their work, and approaching those environments with respect and openness to learn is what I experienced firsthand and will take with me throughout my career.”

de MIT News https://ift.tt/PHrbhn4

MIT sophomores connect with alumni mentors in professional and leadership development program

Team Training Workshop (TTW) is the cornerstone event for students in MIT's Undergraduate Practice Opportunities Program (UPOP), bringing students and alumni mentors together over three exciting sessions for an experiential learning experience that feels like summer camp.

UPOP is a yearlong academic course for MIT sophomores that aids their professional development and helps them prepare for summer internships or other professional experiences — often their first — and for their future careers. UPOP was founded 21 years ago with the recognition that MIT students receive best-in-class technical education, but aren’t given the opportunity to develop the softer skills that will help them succeed in the workplace. 

“I decided to join UPOP because … I want to enter into industry, and I feel like, in addition to the technical skills that I know I'll have at graduation, I also want to build more of those interpersonal skills and team skills with other people around my age as well, who I'll often be working with, and getting advice from people who've already been through all of those different kinds of ins and outs of industry and how to work together,” says Isa Liggans, a current UPOP student majoring electrical engineering and computer science.

Throughout the UPOP program, students engage in skill-building workshops and one-on-one coaching. In the fall, students complete four milestones: resume and cover letter crafting, networking, internship search, and practice interviews. In the spring, they attend workshops on professional communication, project planning, navigating microaggressions, and receiving feedback.

UPOP also connects students with its exclusive employer network, providing them access to a wide range of networking and employment opportunities.

Between semesters is TTW, an intensive experiential learning opportunity that places students in small teams assigned to mentors representing a wide array of industries, most of whom are MIT alums. Teams work together on a series of activities focused on building the skills students will need in the workplace, regardless of what their MIT course is. TTW’s unique program immerses sophomores in professional development, while still prioritizing camaraderie and fun.

The UPOP staff are supported by a dedicated group of mentors, most of whom are MIT alumni, during the fall and spring milestones and TTW. Mentors invest significant time and energy to support UPOP students, and many have been returning to TTW for years, traveling to campus from faraway locations.

“My positive experiences … as a UPOP student, plus the coaching support I received from UPOP staff that helped me secure future internships and even full-time roles, meant that it only seemed natural to want to give back as a mentor,” says Molly Tracy ’16, a mentor and 2016 UPOP alumna. “I continue to find it very gratifying to see students grow in skills that I consistently see as being the differentiator for success in the workplace. This is especially the case for me as a Milestone Mentor. I get to see students I’ve worked with over the course of the fall semester flourish at TTW. These students get clarity on what they want to pursue in their careers, take their first steps to networking, or gain confidence to work in a team. I can’t wait to come back next year!”

Mentors serve as leaders on their assigned teams, providing advice and guidance for students on workshop activities and beyond. They also hold roundtable discussions on a variety of topics, such as negotiating job offers, staying true to your values in your chosen career, what it’s like to work in specific industries, and getting ahead early in one’s career. 

“Working with a mentor is a more practical approach to thinking about teamwork and collaboration,” says Aaliya Hussain, a current UPOP student who is majoring in management. “And I think that one thing that my mentor has helped with, by facilitating the whole table discussions, is just to allow us to connect as a team a little bit better, and get to know each other and work better with each other. And I think another thing is that it's also very interesting to hear the perspective of somebody who's very seasoned at working in different teams.”

TTW is comprised of several training modules — mostly led by MIT faculty and alumni — and culminates in a team presentation.

The Skyscraper module kicks off the event, allowing students to get to know their teammates and compete to build the tallest viable structure out of foam, pencils, and tape. From there, students learn about how to build an effective team, how different thinking and learning styles affect team dynamics and communications, and go through a negotiations exercise where students are assigned the role of an employee pivotal to the future path of a company going through an evolutionary change.

They then learn about the importance of creating consensus while working on technical specifications for a new app, go through rounds of practicing their elevator pitches as part of a UPOP staff-led networking presentation, and receive guidance on making effective presentations in an engaging session with coach and former professional actor Peter Bubriski.

“This was a great experience for me, because in the future, I want to be building high-performance teams who work on really cool projects and who build really cool products,” says Eric Shen, a current UPOP student who is majoring in artificial intelligence and decision making. “And so I guess this was a little preview for me, in terms of where I get to practice communicating with different members, identifying the strengths of each member, and then allocating different tasks that best suits them. And then when you're working in a team, it's not about how one individual performs, but how everybody collectively performs to accomplish the bigger mission. So, it was really fun for me to do that.”

The tips and skills students learn and develop throughout these modules prepare them for the team project announced at the start of TTW. Groups are tasked with creating a sustainability plan for a new green dorm on MIT’s campus, and must meet specific stakeholder needs while staying on budget.

Each day, teams are given an allotted time to work on the project, which they present to a panel of mentors serving as judges on the final day. The goal is for students to consider the facts they are presented with and come up with the most effective solution possible, utilizing newly honed skills in teamwork, creative problem-solving, and communicating effectively.

“I believe that the most important part of it is that you get to try it and do it and then reflect on this experience and what you learned from it,” says Sasha Horokh, a current UPOP student who is majoring in mathematics. “Because it's one thing to think you know how to do that, but another thing to actually do it and realize how to.”

UPOP hosts an employer networking event after the second and third sessions of TTW, giving students an additional opportunity to practice the skills they learned over several days of training modules and team projects. Not only are current UPOP students able to attend the event, but all UPOP alumni (current juniors and seniors) are invited as well. Dozens of employers, including many MIT alumni, attend, giving students a chance to network with people in their desired company or industry.

TTW is as intensive as it is valuable for both MIT sophomores and alumni. It boosts resumes, fosters connections, and gives students a learning experience they wouldn’t have outside UPOP.

As one mentor recalled a member of her team saying, “I came to TTW trying to figure out what type of engineer I wanted to be. I left wanting to figure out what type of person I wanted to be.”

de MIT News https://ift.tt/NhZlq5X

Strengthening trust in machine-learning models

Probabilistic machine learning methods are becoming increasingly powerful tools in data analysis, informing a range of critical decisions across disciplines and applications, from forecasting election results to predicting the impact of microloans on addressing poverty.

This class of methods uses sophisticated concepts from probability theory to handle uncertainty in decision-making. But the math is only one piece of the puzzle in determining their accuracy and effectiveness. In a typical data analysis, researchers make many subjective choices, or potentially introduce human error, that must also be assessed in order to cultivate users’ trust in the quality of decisions based on these methods.

To address this issue, MIT computer scientist Tamara Broderick, associate professor in the Department of Electrical Engineering and Computer Science (EECS) and a member of the Laboratory for Information and Decision Systems (LIDS), and a team of researchers have developed a classification system — a “taxonomy of trust” — that defines where trust might break down in a data analysis and identifies strategies to strengthen trust at each step. The other researchers on the project are Professor Anna Smith at the University of Kentucky, professors Tian Zheng and Andrew Gelman at Columbia University, and Professor Rachael Meager at the London School of Economics. The team’s hope is to highlight concerns that are already well-studied and those that need more attention.

In their paper, published in February in Science Advances, the researchers begin by detailing the steps in the data analysis process where trust might break down: Analysts make choices about what data to collect and which models, or mathematical representations, most closely mirror the real-life problem or question they are aiming to answer. They select algorithms to fit the model and use code to run those algorithms. Each of these steps poses unique challenges around building trust. Some components can be checked for accuracy in measurable ways. “Does my code have bugs?”, for example, is a question that can be tested against objective criteria. Other times, problems are more subjective, with no clear-cut answers; analysts are confronted with numerous strategies to gather data and decide whether a model reflects the real world.

“What I think is nice about making this taxonomy, is that it really highlights where people are focusing. I think a lot of research naturally focuses on this level of ‘are my algorithms solving a particular mathematical problem?’ in part because it’s very objective, even if it’s a hard problem,” Broderick says.

“I think it's really hard to answer ‘is it reasonable to mathematize an important applied problem in a certain way?’ because it's somehow getting into a harder space, it's not just a mathematical problem anymore.”

Capturing real life in a model

The researchers’ work in categorizing where trust breaks down, though it may seem abstract, is rooted in real-world application.

Meager, a co-author on the paper, analyzed whether microfinances can have a positive effect in a community. The project became a case study for where trust could break down, and ways to reduce this risk.

At first look, measuring the impact of microfinancing might seem like a straightforward endeavor. But like any analysis, researchers meet challenges at each step in the process that can affect trust in the outcome. Microfinancing — in which individuals or small businesses receive small loans and other financial services in lieu of conventional banking — can offer different services, depending on the program. For the analysis, Meager gathered datasets from microfinance programs in countries across the globe, including in Mexico, Mongolia, Bosnia, and the Philippines.

When combining conspicuously distinct datasets, in this case from multiple countries and across different cultures and geographies, researchers must evaluate whether specific case studies can reflect broader trends. It is also important to contextualize the data on hand. For example, in rural Mexico, owning goats may be counted as an investment.

“It's hard to measure the quality of life of an individual. People measure things like, ‘What's the business profit of the small business?’ Or ‘What's the consumption level of a household?’ There’s this potential for mismatch between what you ultimately really care about, and what you're measuring,” Broderick says. “Before we get to the mathematical level, what data and what assumptions are we leaning on?”

With data on hand, analysts must define the real-world questions they seek to answer. In the case of evaluating the benefits of microfinancing, analysts must define what they consider a positive outcome. It is standard in economics, for example, to measure the average financial gain per business in communities where a microfinance program is introduced. But reporting an average might suggest a net positive effect even if only a few (or even one) person benefited, instead of the community as a whole.

“What you really wanted was that a lot of people are benefiting,” Broderick says. “It sounds simple. Why didn’t we measure the thing that we cared about? But I think it’s really common that practitioners use standard machine learning tools, for a lot of reasons. And these tools might report a proxy that doesn’t always agree with the quantity of interest.”

Analysts may consciously or subconsciously favor models they are familiar with, especially after investing a great deal of time learning their ins and outs. “Someone might be hesitant to try a nonstandard method because they might be less certain they will use it correctly. Or peer review might favor certain familiar methods, even if a researcher might like to use nonstandard methods,” Broderick says. “There are a lot of reasons, sociologically. But this can be a concern for trust.”

Final step, checking the code 

While distilling a real-life problem into a model can be a big-picture, amorphous problem, checking the code that runs an algorithm can feel “prosaic,” Broderick says. But it is another potentially overlooked area where trust can be strengthened.

In some cases, checking a coding pipeline that executes an algorithm might be considered outside the purview of an analyst’s job, especially when there is the option to use standard software packages.

One way to catch bugs is to test whether code is reproducible. Depending on the field, however, sharing code alongside published work is not always a requirement or the norm. As models increase in complexity over time, it becomes harder to recreate code from scratch. Reproducing a model becomes difficult or even impossible.

“Let’s just start with every journal requiring you to release your code. Maybe it doesn’t get totally double-checked, and everything isn’t absolutely perfect, but let’s start there,” Broderick says, as one step toward building trust.

Paper co-author Gelman worked on an analysis that forecast the 2020 U.S. presidential election using state and national polls in real-time. The team published daily updates in The Economist magazine, while also publishing their code online for anyone to download and run themselves. Throughout the season, outsiders pointed out both bugs and conceptual problems in the model, ultimately contributing to a stronger analysis.

The researchers acknowledge that while there is no single solution to create a perfect model, analysts and scientists have the opportunity to reinforce trust at nearly every turn.

“I don't think we expect any of these things to be perfect,” Broderick says, “but I think we can expect them to be better or to be as good as possible.”

de MIT News https://ift.tt/Fu5GzUW

New additives could turn concrete into an effective carbon sink

Despite the many advantages of concrete as a modern construction material, including its high strength, low cost, and ease of manufacture, its production currently accounts for approximately 8 percent of global carbon dioxide emissions.

Recent discoveries by a team at MIT have revealed that introducing new materials into existing concrete manufacturing processes could significantly reduce this carbon footprint, without altering concrete’s bulk mechanical properties.

The findings are published today in the journal PNAS Nexus, in a paper by MIT professors of civil and environmental engineering Admir Masic and Franz-Josef Ulm, MIT postdoc Damian Stefaniuk and doctoral student Marcin Hajduczek, and James Weaver from Harvard University’s Wyss Institute.

After water, concrete is the world’s second most consumed material, and represents the cornerstone of modern infrastructure. During its manufacturing, however, large quantities of carbon dioxide are released, both as a chemical byproduct of cement production and in the energy required to fuel these reactions. 

Approximately half of the emissions associated with concrete production come from the burning of fossil fuels such as oil and natural gas, which are used to heat up a mix of limestone and clay that ultimately becomes the familiar gray powder known as ordinary Portland cement (OPC). While the energy required for this heating process could eventually be substituted with electricity generated from renewable solar or wind sources, the other half of the emissions is inherent in the material itself: As the mineral mix is heated to temperatures above 1,400 degrees Celsius (2,552 degrees Fahrenheit), it undergoes a chemical transformation from calcium carbonate and clay to a mixture of clinker (consisting primarily of calcium silicates) and carbon dioxide — with the latter escaping into the air.

When OPC is mixed with water, sand, and gravel material during the production of concrete, it becomes highly alkaline, creating a seemingly ideal environment for the sequestration and long-term storage of carbon dioxide in the form of carbonate materials (a process known as carbonation). Despite this potential of concrete to naturally absorb carbon dioxide from the atmosphere, when these reactions normally occur, mainly within cured concrete, they can both weaken the material and lower the internal alkalinity, which accelerates the corrosion of the reinforcing rebar. These processes ultimately destroy the load-bearing capacity of the building and negatively impact its long-term mechanical performance. As such, these slow late-stage carbonation reactions, which can occur over timescales of decades, have long been recognized as undesirable pathways that accelerate concrete deterioration.

“The problem with these postcuring carbonation reactions,” Masic says, “is that you disrupt the structure and chemistry of the cementing matrix that is very effective in preventing steel corrosion, which leads to degradation.”

In contrast, the new carbon dioxide sequestration pathways discovered by the authors rely on the very early formation of carbonates during concrete mixing and pouring, before the material sets, which might largely eliminate the detrimental effects of carbon dioxide uptake after the material cures. 

The key to the new process is the addition of one simple, inexpensive ingredient: sodium bicarbonate, otherwise known as baking soda. In lab tests using sodium bicarbonate substitution, the team demonstrated that up to 15 percent of the total amount of carbon dioxide associated with cement production could be mineralized during these early stages — enough to potentially make a significant dent in the material’s global carbon footprint.

"It's all very exciting," Masic says, "because our research advances the concept of multifunctional concrete by incorporating the added benefits of carbon dioxide mineralization during production and casting.”

Furthermore, the resulting concrete sets much more quickly via the formation of a previously undescribed composite phase, without impacting its mechanical performance. This process thus allows the construction industry to be more productive: Form works can be removed earlier, reducing the time required to complete a bridge or building.

The composite, a mix of calcium carbonate and calcium silicon hydrate, “is an entirely new material,” Masic says. “Furthermore, through its formation, we can double the mechanical performance of the early-stage concrete.” However, he adds, this research is still an ongoing effort. “While it is currently unclear how the formation of these new phases will impact the long-term performance of concrete, these new discoveries suggest an optimistic future for the development of carbon neutral construction materials.”

While the idea of early-stage concrete carbonation is not new, and there are several existing companies that are currently exploring this approach to facilitate carbon dioxide uptake after concrete is cast into its desired shape, the current discoveries by the MIT team highlight the fact that the precuring capacity of concrete to sequester carbon dioxide has been largely underestimated and underutilized.

“Our new discovery could further be combined with other recent innovations in the development of lower carbon footprint concrete admixtures to provide much greener, and even carbon-negative construction materials for the built environment, turning concrete from being a problem to a part of a solution,” Masic says.

The research was supported by the Concrete Sustainability Hub at MIT, which has sponsorship from the Portland Cement Association and the Concrete Research and Education Foundation.

de MIT News https://ift.tt/1EpRlrK

A portfolio that’s out of this world

At age 9, Ezinne Uzo-Okoro SM ’20, PhD ’22 was preoccupied with down-to-earth problems, such as devising an alternative to her father’s messy, paper Filofax organizer, and fixing the unreliable electric service plaguing her home of Owerri, Nigeria. Could she have imagined a path-breaking, 17-year career at NASA, followed by a position as the nation’s space policy expert?

“Absolutely not,” says Uzo-Okoro. “I knew nothing about space — I wanted to be an inventor.”

While she didn’t start as a stargazer, Uzo-Okoro leveraged her curiosity, relationships, voracious appetite for work, and impatience with barriers through a journey that brought her to the center of space exploration, and now to one of the nation’s top science and technology posts as the assistant director of space policy in the White House’s Office of Science and Technology Policy. She began her career at NASA in 2004, where she spent the next 17 years building her expertise in space engineering systems and management. Along the way, she picked up three master’s degrees: one in systems engineering from Johns Hopkins University, one in space robotics from the MIT Media Lab; and in one in public administration from Harvard University. Then in 2022, Uzo-Okoro became the first, and to date only, Black woman to earn a doctorate in aeronautics and astronautics from MIT.

In 2021, Uzo-Okoro began her current position setting the nation’s priorities in space — a sprawling portfolio. On a given day, she might be dealing with the increased proliferation and threat of space debris, crewed and robotic space missions, monitoring the Earth’s climate and space weather, or the International Space Station’s retirement in seven years. It is a kaleidoscopic enterprise driven by innovation benefiting society and the global economy, and one that suits Uzo-Okoro. “This is the best job I’ve ever had,” she says.

Factories in orbit

In April 2022, after Uzo-Okoro convened experts across federal departments and agencies, the White House released a national space policy that addressed an area of burgeoning interest: the use of technologies, including robots, to make and assemble things in space.

Uzo-Okoro is responding to the rising demand among commercial, scientific, and security organizations for satellites that can be customized or manufactured quickly and cost-effectively. It takes months to develop and construct space hardware on the ground, and even longer to ensure the technology will survive on a bone-jarring rocket ride to space.

Setting up orbital factories could dramatically reduce development time and cost for satellites with the ability to sense and monitor natural or human-made disasters. The on-orbit facilities would grow an infrastructure for larger-scale space manufacturing capability, whether for research outposts and habitats on the moon, asteroid mining ventures, or missions to Mars. For all these reasons, “we need to master in-space servicing, assembly, and manufacturing,” says Uzo-Okoro.

Uzo-Okoro first began thinking about the question of space-based manufacturing after years developing small and large spacecraft with NASA. She negotiated time away from the agency to work on the problem — a move inspired by Kerri Cahoy, associate professor in the Department of Aeronautical and Astronautical Engineering (AeroAstro) at MIT, who envisioned producing spacecraft as if they were commodities such as cars. When Uzo-Okoro landed at MIT and began to pursue this idea, Cahoy advised her master’s and doctoral studies.

“Ezinne had this vision of creating a kind of automated factory on orbit, much like those on Earth that use robots to put things together,” says Cahoy. “Her approach was ‘Let’s imagine the future in space where we build important technology, and find the best way to do that.’”

For her master’s and doctoral research, Uzo-Okoro says she was basically “trying to invent the equivalent of an Amazon locker in space” — in essence, a spacecraft in orbit resembling a small refrigerator full of parts, with robot arms to put the parts together. “Inside the locker, you’ve got components for a small satellite like cameras, and spectrometers, and the robot grabs and assembles what you need, rather than creating and assembling on Earth and then launching it.”

Uzo-Okoro mocked up several versions of this robotic space locker, starting on a laboratory workbench and moving to microgravity tests on zero-G flights. “Ezinne came up with the concept, pulled a team together to test it out on a relatively limited budget, and overcame multiple challenges to make it happen — something she’s gotten very good at in her life,” says Cahoy.

Today, a new generation of student researchers plan to take the idea to the next level, with an improved, and larger, locker design. “My work proved that we could assemble a robot autonomously, rather than through human assembly,” Uzo-Okoro says. “The next step is actually putting one of these systems in space.”

A sequence of missions

Through her academic and aerospace careers, Uzo-Okoro has become the inventor of her childhood ambitions. When she left Nigeria to study computer science at Rensselaer Polytechnic Institute (RPI), she “sought the future of technology, where you could literally create anything by learning how to program.” She made first contact with NASA at an RPI job fair. “They told me they were just conducting outreach and not accepting resumes, and I told them that made no sense at all when there was a 30-minute line of talented engineers just waiting to be hired,” she laughs.

That moment served as liftoff for Uzo-Okoro. After graduation, she was hired by the Goddard Space Flight Center, and on her first day, July 12, 2004, the Cassini spacecraft inserted into Saturn’s orbit. On her first assignment, Uzo-Okoro wrote algorithms to help mission physicists identify methane, hydrogen, and nitrogen signals in the data coming home. At NASA, she worked on a series of missions (Earth observation, astrophysics, exoplanet detection, and neutron star interior composition), where she was compelled to devise innovative solutions at every turn.

“I felt like a kid in a candy store, because no matter what I did at NASA, there was always someone who knew more who could teach me,” she says. “When I realized I wanted to be the engineer responsible for a mission, I began educating myself about all parts of the spacecraft design and mission execution.” She studied mechanical and electrical engineering, and began developing and managing entire missions. At NASA, Uzo-Okoro led a small spacecraft mission design center, and was program executive of the heliophysics division.

Uzo-Okoro has not navigated her singular career without meeting obstacles. Being “first” and “only” has left its marks. “You don’t do anything difficult by working 40 hours a week, right?” she notes. She agonized about starting a family — which she ultimately did, while conducting her doctoral research at MIT.

But how did she, as a Black woman, not only survive but prosper in the notably white, notably male aerospace world? “I decided that if you’re just brimming with ideas, get help, particularly if others put up obstacles.” Uzo-Okoro doesn’t trumpet the fact that she’s the first woman to run civilian space policy for the White House. “It’s not important becoming the ‘first’ or ‘only’ one” she says. “The value any of us will bring is results.”

Uzo-Okoro is aware of the responsibility to be a role model, and is fine with leading by example. “If people try something because I’ve done it, that’s great,” she says. “I just keep putting one foot in front of the other and I encourage everybody else to do the same.”

de MIT News https://ift.tt/lAjb1GM