Bridging the Gap Between Safety and Real-Time Performance for Autonomous Vehicle Control

Explore a comprehensive lecture on autonomous vehicle control that bridges the gap between safety and real-time performance. Delve into the Reachability-based Trajectory Design (RTD) method, which combines offline Forward Reachable Set (FRS) computation with real-time trajectory planning. Learn how RTD ensures safety and dynamic feasibility while maintaining efficient planning speed. Discover the method's approach to obstacle representation, persistent feasibility, and its application in various simulations and hardware demonstrations. Gain insights into improving vehicle safety, planning hierarchies, and verification methods such as the Funnel Library. Examine practical examples, including trajectory parameterization, tracking error bounding, and experiments with CarSim full powertrain. Understand how RTD addresses challenges of limited sensor horizons and unpredictable environments in autonomous vehicle operation.

Intro
Improving Vehicle Safety
Why Apply A Planning Hierarchy?
Verification via Select Methods - Funnel Library
An Example of Trajectory Parameterization
Bounding Tracking Error
The Forward Reachable Set (FRS)
Goal for Online Computation
CarSim Full Powertrain Experiment
Moving Obstacles