Driving · 2026-10-30
Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation
Jacob Levy, Arthur Bouton, Tristan D. Hassler, Hari Nayar, David Fridovich-Keil
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
TERRA is an online adaptation framework for planetary rovers that adapts a simulation-pretrained world model using a reward-weighted prediction loss and trains a loss predictor to target where future performance-relevant model errors are likely. It uses this signal to guide exploration during MPPI-based MPC. In simulation it reaches ~90% slope-climbing success within five trials under severe pretraining shifts, and on the ERNEST rover it recovers coordinated AGS locomotion within three to five real-world trials.
Why it matters
The authors target sim-to-real gaps in deformable, task-relevant rover dynamics, where unlimited trial-and-error is impractical on deployment hardware. By prioritizing reward-relevant model errors for adaptation and guiding data collection toward interactions that matter for MPC returns, the method aims to achieve efficient deployment-time adaptation on out-of-distribution deformable terrain.
Method
- Adaptation: keep transformer backbone fixed, finetune only lightweight LoRA parameters in the dynamics model using a reward-weighted prediction loss. Updates are regularized to anchor toward simulation dynamics.
- Loss predictor: train an online transformer that predicts discretized, high-loss bins over the same candidate action sequences used by MPC, using class-weighted cross-entropy.
- Exploration-MPC loop: convert loss-predictor logits into an exploration bonus added to the MPPI scoring objective, execute the first action, then update dynamics and loss predictor with newly collected deployment transitions.
Abstract (from arXiv)
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
Related papers
- LocoWM: High-Precision Locomotion through World-Model-Guided Residual Adaptation
- TerraTransfer++: Demonstration-Free End-to-End Planner
- From Pixels to Executable Physics: Behavior-Grounded World Models for Embodied Action Consequence Prediction
- LynxScenes: A Large-Scale Real-World Benchmark for Outdoor Wheel-Legged Robot Perception, Navigation, Imagination and Reconstruction
- T2-z: Test-Time Optimization of Environment Latents for Robotic World Models
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First announced by arXiv: 2026-10-30 · Indexed by Pub-AI: 2026-10-30 UTC
Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation
First announced by arXiv: 2026-10-30 · Original record ↗
- arxiv
- 2610.38657
Explicit source versions (1)
- v1 · Date not supplied
- Preprint available
Source date: 2026-10-30 · First detected: 2026-10-30 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · RSS guid=oai:arXiv.org:2610.38657v1; arxiv:announce_type=new; item pubDate (official metadata)
Abstract · arxiv
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