Wang on Planning, Operation, and Management of Automated Transportation Systems

Shi'an Wang, a Ph.D. student advised by Michael Levin, completed doctoral requirements on Wednesday, November 16, 2022. In the spring, Wang will join the faculty at the University of Texas at El Paso at the rank of Assistant Professor. 
Wang's thesis addressed "Planning, Operation, and Management of Automated Transportation Systems: A Control-Theoretic Approach."
Automated vehicles (AVs) are expected to become increasingly available, offering benefits such as improved roadway safety, enhanced traffic stability, and reduced energy consumption, among others. A transitional period in the auto market is anticipated as human-driven vehicles are gradually replaced by AVs. Many opportunities and challenges will emerge during this transitional period. Wang's doctoral research addresses some of the interesting and pressing problems arising from vehicle automation in the context of planning, operation, and management of future transportation systems from a control-theoretic perspective.
Wang first addresses the problem of designing optimal incentive programs for accelerating adoption of AVs. Decision variables, such as subsidies for AV purchases and investment in AV infrastructure, are considered as ways to promote the adoption of AVs and achieve a desired temporal integration of them into the auto market. Wang formulates this as an optimal control problem, designed to attain a desired market penetration rate (MPR) of AVs while minimizing the costs of subsidies and infrastructure.
While incentive policies could accelerate the adoption of AVs, the MPR is expected to remain low for the next few decades, resulting in a predominantly human-driven mixed traffic flow. Such flow has been shown to be unstable in certain regimes due to collective behavior of human drivers causing stop-and-go waves. Emerging AV technologies open a door for mitigating these undesired traffic waves with AVs acting as mobile actuators (known as Lagrangian traffic control). Wang describes design of optimal AV controls with only a small proportion of AVs present, resulting in smoother traffic flow and reduced energy consumption. The proposed approach is effective in traffic smoothing; however, it is inadequate to ensure traffic stability. 
Hence, Wang proposes a new idea of virtual tracking,  where a carefully constructed virtual speed profile is closely tracked by an AV. Wang shows that a class of synthesized AV controllers is effective in smoothing nonlinear mixed traffic, with analytical guarantees on the convergence of speed tracking and car-following safety.
Finally, Wang envisions a futuristic scenario in which a select number of AVs are compromised by cyberattacks to drive in an adversarial manner, degrading the performance of the transportation systems. To deal with the lack of knowledge about malicious attacks, Wang formulated a min-max control problem to minimize the worst-case potential disturbance to traffic flow. Following the set of necessary conditions of optimality derived for the formulated optimization problem, an iterative computational algorithm is developed for determining the optimal control (driving) strategy of AVs in the presence of attacks.
Wang's work will help the nation be better prepared for the arrival of automated vehicles and help protect the people who drive or ride within our transportation systems.