Rovers adapt quickly to rough terrain with smart targeted exploration
Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation
RoboticsArtificial IntelligenceMachine Learning
Summary
Rovers exploring other planets often struggle because the ground behaves in unexpected ways that models can’t fully predict. The authors developed a method called TERRA that helps rovers learn from a few tests while they operate, focusing on the most important mistakes for completing their tasks. This lets rovers quickly improve and succeed on tricky slopes without wasting time on random trial and error. They tested TERRA in simulations and on a real Mars-like robot, which adapted well to difficult terrain in just a few tries.
What this means in practice
- •For robotics system developers: Enable planetary rovers to adapt their mobility strategies on steep and soft terrain by focusing exploration on critical model errors during deployment.
- •For autonomous vehicle engineers: Improve off-road vehicle performance by online correcting simulation-based models with targeted exploration of terrain-related errors in real environments.
Authors
Jacob Levy, Arthur Bouton, Tristan D. Hassler, Hari Nayar, David Fridovich-Keil
Abstract
Future planetary rovers will traverse steep, deformable terrain where slip, sinkage, and terrain deformation can threaten mobility. Although simulation can capture some wheel-soil effects, unmodeled deformable-terrain phenomena create sim-to-real gaps requiring deployment-time adaptation. Yet, rover operations leave little room for extensive trial-and-error. We introduce TERRA (Targeted Exploration for Reward-Relevant Adaptation), an online adaptation framework that efficiently seeks deployment data exposing task-relevant model errors and corrects them. TERRA adapts a simulation-trained world model with a reward-weighted prediction loss and trains a loss predictor to estimate where future performance-relevant errors are likely to occur, using this signal to guide exploration during MPC. In simulation, TERRA reaches approximately 90% slope-climbing success within five trials under severe out-of-distribution pretraining conditions, outperforming exploration baselines in reward, completion time, and energy to first success. On the ERNEST rover, TERRA recovers coordinated active-suspension locomotion on a deformable Mars-simulant slope within three to five real-world trials. Project website: https://terra-explore.github.io