arXiv cs.AI· Jacob Levy, Arthur Bouton, Tristan D. Hassler, Hari Nayar, David Fridovich-Keil·· 3 小时前AI 评分36
TERRA：面向火星车地形自适应的定向探索框架
Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation
AI 导读
研究者提出 TERRA（Targeted Exploration for Reward-Relevant Adaptation），一个在线自适应框架，通过奖励加权预测损失微调仿真训练的世界模型，并用损失预测器引导 MPC 期间的定向探索，快速修正部署时未建模的可变形地形误差。
正文 · 原文
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: this https URL
| Comments: | Conference on Robot Learning 2026 |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.38657 [cs.RO] |
| (or arXiv:2610.38657v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.38657 arXiv-issued DOI via DataCite |
Submission history
From: Jacob Levy [view email]
[v1]
Wed, 28 Oct 2026 18:41:55 UTC (4,406 KB)
来源：arXiv cs.AI · arxiv.org