Safe, High-performance Motion Planning Under Uncertainty for Autonomous Driving Applications

Jacob Knaup | Ph.D. Student, Institute for Robotics and Intelligent Machines at Georgia Institute of Technology 
Monday November 18 @ 1:30pm in Montgomery Knight 317 

Abstract: Autonomous driving is a high-performance, safety-critical task, wherein a robot must balance the trade-off between driving aggressively to meet its objectives while maintaining a suitable level of safety to avoid crashes. This challenge is further exacerbated by the necessity of maintaining this balance while dealing with uncertainty in the environment, whether from uncertainty in the autonomous vehicle’s own dynamics or from uncertainty in other vehicles’ behaviors. This dissertation will present theoretical and experimental results from two novel methods for motion planning for uncertain systems that integrate ideas from the model-based and learning-based stochastic optimal control communities. The first utilizes a convex programming-based stochastic model-predictive control method for aggressive off-road autonomous racing. The second addresses a highway merge scenario and employs active learning of other drivers’ behaviors and model-based generative diffusion for interactive motion planning.

Thesis committee:
Dr. Panagiotis Tsiotras
The Daniel Guggenheim School of Aerospace Engineering
Georgia Institute of Technology

Dr. Samuel Coogan
School of Electrical and Computer Engineering
Georgia Institute of Technology

Dr. Kyriakos Vamvoudakis
The Daniel Guggenheim School of Aerospace Engineering
Georgia Institute of Technology

Dr. Ye Zhao
George W. Woodruff School of Mechanical Engineering
Georgia Institute of Technology

Dr. Francesco Borrelli
Department of Mechanical Engineering
University of California, Berkeley