By Grace Chen and Vishrut Rana

Consider this: you are driving back from a long day at work but instead of being halted by the endless stretches of traffic and entering the infinite cycle of starting and stopping that you are used to — your drive is free of frustrating congestions and you make it home in half the time. How beautiful would this world be? Under the leadership of Professor Alexandre Bayen, researchers from UC Berkeley’s Institute for Transportation Studies on the CIRCLES project (Congestion Impact Reduction via CAV-in-the-loop Lagrangian Energy Smoothing) see this world as more than a mere imagination. The CIRCLES project’s goal is to use autonomous vehicles to improve the traffic flow of manned vehicles. CIRCLES works by using FLOW, a software framework that simulates scenarios to control traffic using principles of machine learning. The framework tests common traffic tasks like controlling highway merges and bottlenecks to study common problems and identify how they can be remedied. The key impacts of the CIRCLES project are threefold: effectively controlling traffic, minimizing vehicular emissions, and making roads safer.

In order to begin testing different methods to improve these issues, the team first created a sample of the concept. This simplified sample served as proof that the premise of the initiative was possible and could work in more complex scenarios to smooth traffic flow. As the team began developing more complex computer models of traffic scenarios, they sought a physical testing environment for their work. With the help of graduate student researcher Kathy Jang, the team chose to collaborate with the University of Delaware Scaled Smart City (UDSSC). UDSSC is a model of a smart city with 35 robotic vehicles that can help emulate real-world traffic scenarios in a small and controlled environment. This is an ideal testbed for CIRCLES to prove concepts beyond the simulation level and understand the impacts of hardware errors and delays. By employing test conditions to UDSCC, the research team is able to test their computer models and illustrate how autonomous vehicles could help maintain traffic flow in their envisioned future “smart city.”

To understand how the concept of CIRCLES works, it is essential to understand one of the root causes of traffic. Contrary to popular belief, a common source of traffic isn’t one particularly slow car holding back the whole lane. The cause may be all cars driving as greedily as possible by aggressively tailgating and speeding. This in turn causes small and random braking events that are amplified to system-wide stop-and-go waves. However, autonomous vehicles have the ability to drive with social utility in mind and communicate with other autonomous vehicles. So, if autonomous vehicles were integrated with other manned vehicles, all traffic on the road would be controlled to follow the optimal speed of autonomous vehicles. Hence, by following the speed of the autonomous vehicles, manned vehicles would be able to drive at a more consistent speed and avoid congestion caused by stop-and-go waves. This would be especially valuable to achieve optimal traffic flow in common traffic situations such as mergers, roundabouts, exits, and intersections.

CIRCLES, however, does not come without its limitations. The main barrier associated with the project is the difference in the simulations from reality, which Jang calls the “reality gap.” Explaining the reality gap, graduate student researcher Fangyu Wu stated “We assume cars act in one way in our clean model; but in the real world, they act differently. So, there is going to be an error.” Furthering his explanation, Wu continued “say you can play a music piece very well on a specific violin. But if someone gives you a new violin, you might struggle because it is different and you need to adjust.” The researchers aim to build a framework that is successful in every type of environment, like the unpredictable physical world, and not just the simulated environment of UDSSC.

Nevertheless, the researchers remain positive regarding their ability to minimize this reality gap. And when successful, the CIRCLES project would allow manned vehicles to drive more efficiently and therefore reduce their associated emissions. For instance, autonomous vehicles could help reduce the repeated use of the gas and brake pedals — a major source of decreased fuel efficiency in cars. Such stop-and-go instances mostly occur at intersections, exits, and mergers, which are areas of focus for CIRCLES already. As a result, CIRCLES could ultimately reduce carbon emissions from vehicles. In addition to making transportation more sustainable, CIRCLES can also improve safety on the roads. To achieve this, the project team is researching ways that autonomous vehicles can communicate with each other to reduce the chance of a collision. Professor Bayen hopes to “revolutionize traffic management” and envisions a future where “autonomous vehicles take the place of traffic lights, speed limits, and road signs.” Professor Bayen recognizes that although this might sound like a shocking concept, it is important to remember that “traffic lights used to be a revolutionary idea as well when the amount of cars on the roads first started increasing, but now they are mainstream.” For instance, imagine if two autonomous vehicles on different roads could communicate with each other. When the two roads intersect, their respective traffic flows would not collide because the autonomous vehicles could “talk” about who is going to go first. This would deem traffic lights and other safety measures obsolete, because an autonomous vehicle does not need something to tell it to stop, they can just “talk” to the other vehicles on the road. As a result, CIRCLES would make the drive safer without having to install and maintain expensive infrastructure such as traffic lights and stop signs.

What makes the work done by the researchers all the more interesting is that they are not alone in their endeavor. The team has an open source code for the project (at https://github.com/flow-project/flow), meaning that anyone can access and use the code that the team has written. The researchers aim to build a community around CIRCLES and actively collaborate with other researchers. This illustrates the team’s dedication to the NSPE Code of Ethics for Engineers and specifically their first fundamental canon which reads “hold paramount the safety, health, and welfare of the public.” By creating a community that shares and builds on each other’s knowledge, each research party benefits and gets one step closer to creating more sustainable transportation for the general public. Rather than being motivated by selfish reasons, Professor Bayen’s team prioritizes the welfare of the public by sharing their learnings with the community. (Courtesy of National Society of Professional Engineers Logo) Looking at the larger picture, it is only natural to be intrigued by the implementation timeline for this project. The researchers think that the deployment of this concept at a large scale is still far away due to accountability and legal challenges associated with autonomous cars. With respect to this, graduate student researcher Abdul Kreidieh highlighted that “there are issues with safety and accountability that people are trying to figure out, like who is responsible for what component. For example, if a plane were to crash, there is accountability on multiple people’s parts. The same thing will start happening for autonomous vehicles, but this is still being figured out.” However, the team emphasized that their focus, at this stage, is to refine their technology by controlling traffic only under specific simulated conditions. Until then, the possibility of not being stuck in a never-ending roadblock, and doing so in a safer and more environmentally efficient way, is reason enough to be excited. The work engineers do shapes the world around us. But given the technical nature of that work, non-engineers may not always realize the impact and reach of engineering research. In E185: The Art of STEM Communication, students learn about and practice written and verbal communication skills that can bring the world of engineering to a broader audience. They spend the semester researching projects within the College of Engineering, interviewing professors and graduate students, and ultimately writing about and presenting that work for a general audience. This piece is one of the outcomes of the E185 course. Connect with Grace Chen (BS IEOR ’22) and Vishrut Rana (BS IEOR ’22)

---

Say goodbye to rush hour traffic was originally published in Berkeley Master of Engineering on Medium, where people are continuing the conversation by highlighting and responding to this story.