This research focuses on the challenges faced by autonomous vehicles traversing uneven terrain at high speeds. The study introduces a physics-based framework for identifying traversability constraints and mitigating risks such as rollover and ditch-crossing failures. It presents a planning and control system that uses Model Predictive Control (MPC) and a low-level controller for efficient computation of these constraints. The approach was tested through real-world experimentation and compared against a human expert, demonstrating its effectiveness in safe and aggressive control on uneven terrain.
Publication date: 22 Nov 2023
Project Page: https://sites.google.com/cs.washington.edu/off-road-mpc
Paper: https://arxiv.org/pdf/2311.12284
