Examining the world through signals and systems

Theres a mesmerizing video life on YouTube of simulated self-driving commerce streaming through a six-lane four-way intersection. Dozens of cars flow through the streets pausing turning slowing and speeding up to quit colliding with their neighbors. And not a one car stopping. But what if even one of those vehicles was not autonomous? What if only one was?

In the coming decades autonomous vehicles will play a growing role in aggregation whether care drivers safer making deliveries or increasing accessibility and mobility for elderly or disabled passengers.

But MIT Assistant Professor Cathy Wu argues that autonomous vehicles are just part of a intricate transport method that may implicate personal self-driving cars delivery fleets ethnical drivers and a range of last-mile solutions to get passengers to their doorstep — not to mention road infrastructure like highways roundalmosts and yes intersections.

Transport today accounts for almost one-third of U.S. energy decline. The decisions we make today almost autonomous vehicles could have a big contact on this number — ranging from a 40 percent decrease in energy use to a doubling of energy decline.

So how can we better apprehend the problem of integrating autonomous vehicles into the transportation method? Equally significant how can we use this apprehending to lead us toward better-functioning methods?

Wu who joined the Laboratory for Information and Decision Systems (LIDS) and MIT in 2019 is the Gilbert W. Winslow Assistant Professor of Civil and Environmental Engineering as well as a core faculty limb of the MIT Institute for Data Systems and Society. Growing up in a Philadelphia-area family of electrical engineers Wu sought a field that would empower her to harness engineering skills to explain societal challenges. 

During her years as an undergraduate at MIT she reached out to Professor Seth Teller of the Computer Science and Artificial Intelligence Laboratory to debate her interest in self-driving cars.

Teller who passed away in 2014 met her questions with warm advice says Wu. ’He told me If you have an idea of what your emotion in life is then you have to go behind it as hard as you perhaps can. Only then can you hope to find your true emotion.

’Anyone can tell you to go behind your dreams but his insight was that dreams and ambitions are not always clear from the set. It takes hard work to find and chase your emotion.’ 

Chasing that emotion Wu would go on to work with Teller as well as in Professor Daniela Russ Distributed Robotics Laboratory and finally as a graduate student at the University of California at Berkeley where she won the IEEE Intelligent Transportation Systems Societys best PhD assign in 2019.

In graduate school Wu had an epiphany: She realized that for autonomous vehicles to fulfill their promise of fewer accidents time saved lower emissions and greater socioeconomic and natural accessibility these goals must be explicitly designed-for whether as natural infrastructure algorithms used by vehicles and sensors or deliberate plan decisions.

At LIDS Wu uses a type of machine learning named reinforcement learning to study how commerce methods behave and how autonomous vehicles in those methods ought to behave to get the best practicable outcomes.

Reinforcement learning which was most famously used by AlphaGo DeepMinds ethnical-beating Go program is a strong class of methods that capture the idea behind trial-and-error — given an extrinsic a learning agent frequently attempts to accomplish the extrinsic failing and learning from its mistakes in the process.

In a commerce method the extrinsics might be to maximize the overall mean quickness of vehicles to minimize journey time to minimize energy decline and so on.

When studying ordinary components of commerce networks such as grid roads bottlenecks and on- and off-ramps Wu and her colleagues have establish that reinforcement learning can match and in some cases exceed the accomplishment of running commerce control strategies. And more significantly reinforcement learning can shed new light toward apprehending intricate networked methods — which have long evaded pure control techniques. For entreaty if just 5 to 10 percent of vehicles on the road were autonomous and used reinforcement learning that could cast congeries and boost vehicle speeds by 30 to 140 percent. And the learning from one scenario frequently translates well to others. These insights could one day soon help to enlighten open plan or business decisions.

In the order of this investigation Wu and her colleagues helped better a class of reinforcement learning methods named plan gradient methods. Their advancements turned out to be a general betterment to most existing deep reinforcement learning methods.

But reinforcement learning techniques will need to be constantly betterd to keep up with the layer and shifts in infrastructure and changing conduct patterns. And investigation findings will need to be translated into action by urban planners auto makers and other organizations.

Today Wu is collaborating with open agencies in Taiwan and Indonesia to use insights from her work to lead better dialogues and decisions. By changing commerce signals or using nudges to shift drivers conduct are there other ways to accomplish lower emissions or smoother commerce? 

’Im surprised by this work see day’ says Wu. ’We set out to reply a question almost self-driving cars and it turns out you can pull aloof the insights adduce them in other ways and then this leads to new exciting questions to reply.’

Wu is lucky to have establish her mental home at LIDS. Her experience of it is as a ’very deep mental well-inclined and welcoming locate.’ And she counts among her investigation inspirations MIT order 6.003 (Signals and Systems) — a class she encourages seeone to take — taught in the tradition of professors Alan Oppenheim (Research Laboratory of Electronics) and Alan Willsky (LIDS). ’The order taught me that so much in this globe could be fruitfully examined through the lens of signals and methods be it electronics or institutions or aggregation’ she says. ’I am just realizing as Im assertion this that Ive been empowered by LIDS thinking all along!’

Research and training through a pandemic havent been easy but Wu is making the best of a challenging leading year as faculty. (’Ive been working from home in Cambridge — my brief walking alter is irrelevant at this point’ she says wryly.) To unwind she enjoys running listening to podcasts covering topics ranging from science to history and reverse-engineering her favorite Trader Joes frozen foods.

Shes also been working on two Covid-related projects born at MIT: One explores how data from the environment such as data calm by internet-of-things-connected thermometers can help unite emerging aggregation outbreaks. Another project asks if its practicable to prove how catching the virus is on open transport and how different factors might decrease the transmission risk.

Both are in their soon stages Wu says. ’We hope to conduce a bit to the pool of apprehension that can help decision-makers somewhere. Its been very enlightening and rewarding to do this and see all the other efforts going on about MIT.’