CEE postdoctoral scholar Shuo Feng has won second prize in the IEEE Intelligent Transportation Systems Society (ITSS) Best PhD Dissertation Award for his dissertation, “Testing Scenario Library Generation for Connected and Automated Vehicles.”
Feng’s dissertation develops a library of efficient, scenario-based testing models for connected and automated vehicles (CAVs). Testing library scenario generation (TSLG) is crucial to the advancement of CAVs, as comparing automated vehicles to human-driven vehicles would require hundreds of millions of miles worth of driving tests. It’s important that these often-virtual simulations are varied in their scenarios, both for testing bias and for approximating real-world situations.
Feng conducted much of his dissertation research while a visiting PhD student at U-M from Tsinghua University, noting that his methodological breakthrough occurred during his time in Ann Arbor. He adds that he relied on the human driving behaviors datasets at the University of Michigan Transportation Research Institute (UMTRI), in particular. He is now a postdoc with the Department of Civil and Environmental Engineering (CEE), where he continues to work with UMTRI.
Feng’s dissertation is both theoretical and practical, developing a methodological framework for creating models and implementing them in a testing library at Mcity with real CAVs. “The breakthrough is enabled by (my) discovery of a new scenario evaluation metric, scenario criticality, to ensure that the selected scenarios are both naturalistic and adversarial, and the importance sampling theory can be applied to accelerate the evaluation process,” writes Feng.
The ITSS prize is awarded to the “best dissertation in any ITS area that is innovative and relevant to practice. This award is established to encourage doctoral research that combines theory and practice, makes in-depth technical contributions, or is interdisciplinary in nature,” notes the IEEE ITSS website.