GRASP Industry Talk: Honda Research Institute, “Working with Imperfect Prediction on Autonomous Vehicles”
February 21 @ 10:45 am - 11:45 am
This will be a hybrid event with in-person attendance in Wu and Chen and virtual attendance on Zoom. This seminar will NOT be recorded.
ABSTRACT
State of the art prediction can fail catastrophically when operating outside of the training distribution. We show methods that can be used to reduce these failures. We then outline how delayed decision making can be used to safely plan within the constraints of these block box prediction methods.
Presenter
David Isele - Learn More
David Isele is a principal research scientist at Honda Research Institute in San Jose, California. His research focuses on interactive decision making for autonomous vehicles. This includes game theoretic negotiation strategies, making learning approaches more reliable, combining learning with control theory optimization, and planning with uncertainty.