MIT researchers measure traffic emissions, to the block, in real-time

In a study focused on New York City, MIT researchers have shown that existing sensors and mobile data can be used to generate a near real-time, high-resolution picture of auto emissions, which could be used to develop local transportation and decarbonization policies.

The new method produces much more detailed data than some other common approaches, which use intermittent samples of vehicle emissions. The researchers say it is also more practical and scales up better than some studies that have aimed for very granular emissions data from a small number of automobiles at once. The work helps bridge the gap between less-detailed citywide emissions inventories and highly detailed analyses based on individual vehicles.

“Our model, by combining real-time traffic cameras with multiple data sources, allows extrapolating very detailed emission maps, down to a single road and hour of the day,” says Paolo Santi, a principal research scientist in the MIT Senseable City Lab and co-author of a new paper detailing the project’s results. “Such detailed information can prove very helpful to support decision-making and understand effects of traffic and mobility interventions.”

Carlo Ratti, director of the MIT Senseable City Lab, notes that the research “is part of our lab’s ongoing quest into hyperlocal measurements of air quality and other environmental factors. By integrating multiple streams of data, we can reach a level of precision that was unthinkable just a few years ago — giving policymakers powerful new tools to understand and protect human health.”

The new method also protects privacy, since it uses computer vision techniques to recognize types of vehicles, but without compiling license plate numbers. The study leverages technologies, including those already installed at intersections, to yield richer data about vehicle movement and pollution.

“The very basic idea is just to estimate traffic emissions using existing data sources in a cost-effective way,” says Songhua Hu, a former postdoc in the Senseable City Lab, and now an assistant professor at City University of Hong Kong.

The paper, “Ubiquitous Data-driven Framework for Traffic Emission Estimation and Policy Evaluation,” is published in Nature Sustainability.

The authors are Hu; Santi; Tom Benson, a researcher in the Senseable City Lab; Xuesong Zhou, a professor of transportation engineering at Arizona State University; An Wang, an assistant professor at Hong Kong Polytechnic University; Ashutosh Kumar, a visiting doctoral student at the Senseable City Lab; and Ratti. The MIT Senseable City Lab is part of MIT’s Department of Urban Studies and Planning.

Manhattan measurements

To conduct the study, the researchers used images from 331 cameras already in use in Manhattan intersections, along with anonymized location records from over 1.75 million mobile phones. Applying vehicle-recognition programs and defining 12 broad categories of automobiles, the scholars found they could correctly place 93 percent of vehicles in the right category. The imaging also yielded important information about the specific ways traffic signals affect traffic flow. That matters because traffic signals are a major reason for stop-and-go driving patterns, which strongly affect urban emissions but are often omitted in conventional inventories.

The mobile phone data then provided rich information about the overall patterns of traffic and movement of individual vehicles throughout the city. The scholars combined the camera and phone data with known information about emissions rates to arrive at their own emissions estimates for New York City.

“We just need to input all emission-related information based on existing urban data sources, and we can estimate the traffic emissions,” Hu says.

Moreover, the researchers evaluated the changes in emissions that might occur in different scenarios when traffic patterns, or vehicle types, also change.

For one, they modeled what would happen to emissions if a certain percentage of travel demand shifted from private vehicles to buses. In another scenario, they looked at what would happen if morning and evening rush hour times were spread out a bit longer, leaving fewer vehicles on the road at once. They also modeled the effects of replacing fine-grained emissions inputs with citywide averages — finding that the rougher emissions estimates could vary widely, from −49 percent to 25 percent of the more fine-tuned results. That underscores how seemingly small simplifications can introduce large errors into emission estimates.

Major emissions drop

On one level, this work involved altering inputs into the model and seeing what emerged. But one scenario the researchers studied is based on a real-world change: In January 2025, New York City implemented congestion pricing south of 60th Street in Manhattan.

To study that, the researchers looked at what happened to vehicle traffic at intervals of two, four, six, and eight weeks after the program began. Overall, congestion pricing lowered traffic volume by about 10 percent — but there was a corresponding drop in emissions of 16-22 percent.

This finding aligns with a previous study by researchers at Cornell University, which reported a 22 percent reduction in particulate matter (PM2.5) levels within the pricing zone. The MIT team also found that these reductions were not evenly distributed across the network, with larger declines on some major streets and more mixed effects outside the pricing zone.

“We see these kinds of huge changes after the congestion pricing began, Hu says. “I think that’s a demonstration that our model can be very helpful if a government really wants to know if a new policy converts into real-world impact.”

There are additional forms of data that could be fed into the researchers’ new method. For instance, in related work in Amsterdam, the team leveraged dashboard cams from vehicles to yield rich information about vehicle movement.

“With our model we can make any camera used in cities, from the hundreds of traffic cameras to the thousands of dash cams, a powerful device to estimate traffic emissions in real-time,” says Fábio Duarte, the associate director of research and design at the MIT Senseable City Lab, who has worked on multiple related studies.

The research was supported by the MIT Senseable City Consortium, which consists of Atlas University, the city of Laval, the city of Rio de Janeiro, Consiglio per la Ricerca in Agricoltura e l’Analisi dell’Economia Agraria, the Dubai Future Foundation, FAE Technology, KAIST Center for Advanced Urban Systems, Sondotecnica, Toyota, and Volkswagen Group America.

de MIT News https://ift.tt/6vSt4wA

Evaluating the ethics of autonomous systems

Artificial intelligence is increasingly being used to help optimize decision-making in high-stakes settings. For instance, an autonomous system can identify a power distribution strategy that minimizes costs while keeping voltages stable.

But while these AI-driven outputs may be technically optimal, are they fair? What if a low-cost power distribution strategy leaves disadvantaged neighborhoods more vulnerable to outages than higher-income areas?

To help stakeholders quickly pinpoint potential ethical dilemmas before deployment, MIT researchers developed an automated evaluation method that balances the interplay between measurable outcomes, like cost or reliability, and qualitative or subjective values, such as fairness.   

The system separates objective evaluations from user-defined human values, using a large language model (LLM) as a proxy for humans to capture and incorporate stakeholder preferences. 

The adaptive framework selects the best scenarios for further evaluation, streamlining a process that typically requires costly and time-consuming manual effort. These test cases can show situations where autonomous systems align well with human values, as well as scenarios that unexpectedly fall short of ethical criteria.

“We can insert a lot of rules and guardrails into AI systems, but those safeguards can only prevent the things we can imagine happening. It is not enough to say, ‘Let’s just use AI because it has been trained on this information.’ We wanted to develop a more systematic way to discover the unknown unknowns and have a way to predict them before anything bad happens,” says senior author Chuchu Fan, an associate professor in the MIT Department of Aeronautics and Astronautics (AeroAstro) and a principal investigator in the MIT Laboratory for Information and Decision Systems (LIDS).

Fan is joined on the paper by lead author Anjali Parashar, a mechanical engineering graduate student; Yingke Li, an AeroAstro postdoc; and others at MIT and Saab. The research will be presented at the International Conference on Learning Representations.

Evaluating ethics

In a large system like a power grid, evaluating the ethical alignment of an AI model’s recommendations in a way that considers all objectives is especially difficult.

Most testing frameworks rely on pre-collected data, but labeled data on subjective ethical criteria are often hard to come by. In addition, because ethical values and AI systems are both constantly evolving, static evaluation methods based on written codes or regulatory documents require frequent updates.

Fan and her team approached this problem from a different perspective. Drawing on their prior work evaluating robotic systems, they developed an experimental design framework to identify the most informative scenarios, which human stakeholders would then evaluate more closely.

Their two-part system, called Scalable Experimental Design for System-level Ethical Testing (SEED-SET), incorporates quantitative metrics and ethical criteria. It can identify scenarios that effectively meet measurable requirements and align well with human values, and vice versa.   

“We don’t want to spend all our resources on random evaluations. So, it is very important to guide the framework toward the test cases we care the most about,” Li says.

Importantly, SEED-SET does not need pre-existing evaluation data, and it adapts to multiple objectives.

For instance, a power grid may have several user groups, including a large rural community and a data center. While both groups may want low-cost and reliable power, each group’s priority from an ethical perspective may vary widely.

These ethical criteria may not be well-specified, so they can’t be measured analytically.

The power grid operator wants to find the most cost-effective strategy that best meets the subjective ethical preferences of all stakeholders.

SEED-SET tackles this challenge by splitting the problem into two, following a hierarchical structure. An objective model considers how the system performs on tangible metrics like cost. Then a subjective model that considers stakeholder judgements, like perceived fairness, builds on the objective evaluation.

“The objective part of our approach is tied to the AI system, while the subjective part is tied to the users who are evaluating it. By decomposing the preferences in a hierarchical fashion, we can generate the desired scenarios with fewer evaluations,” Parashar says.

Encoding subjectivity

To perform the subjective assessment, the system uses an LLM as a proxy for human evaluators. The researchers encode the preferences of each user group into a natural language prompt for the model.

The LLM uses these instructions to compare two scenarios, selecting the preferred design based on the ethical criteria.

“After seeing hundreds or thousands of scenarios, a human evaluator can suffer from fatigue and become inconsistent in their evaluations, so we use an LLM-based strategy instead,” Parashar explains.

SEED-SET uses the selected scenario to simulate the overall system (in this case, a power distribution strategy). These simulation results guide its search for the next best candidate scenario to test.

In the end, SEED-SET intelligently selects the most representative scenarios that either meet or are not aligned with objective metrics and ethical criteria. In this way, users can analyze the performance of the AI system and adjust its strategy.

For instance, SEED-SET can pinpoint cases of power distribution that prioritize higher-income areas during periods of peak demand, leaving underprivileged neighborhoods more prone to outages.

To test SEED-SET, the researchers evaluated realistic autonomous systems, like an AI-driven power grid and an urban traffic routing system. They measured how well the generated scenarios aligned with ethical criteria.

The system generated more than twice as many optimal test cases as the baseline strategies in the same amount of time, while uncovering many scenarios other approaches overlooked.

“As we shifted the user preferences, the set of scenarios SEED-SET generated changed drastically. This tells us the evaluation strategy responds well to the preferences of the user,” Parashar says.

To measure how useful SEED-SET would be in practice, the researchers will need to conduct a user study to see if the scenarios it generates help with real decision-making.

In addition to running such a study, the researchers plan to explore the use of more efficient models that can scale up to larger problems with more criteria, such as evaluating LLM decision-making.

This research was funded, in part, by the U.S. Defense Advanced Research Projects Agency.

de MIT News https://ift.tt/uo4xFUj