Learning Locomotion on Complex Terrain for Quadrupedal Robots with Foot Position Maps and Stability Rewards

arXiv cs.RO / 4/6/2026

💬 OpinionIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper tackles quadrupedal robot locomotion on complex terrain by explicitly incorporating a foot position map integrated with a heightmap, rather than inferring foot placement implicitly from joint angles.
It proposes a dynamic locomotion-and-stability reward within an attention-based framework to provide stronger stability guidance than prior RL approaches.
The method is validated both on terrains encountered during training (in-domain) and on out-of-domain (OOD) terrains to assess generalization.
Experimental results indicate improved locomotion precision and stability, translating into higher locomotion success rates across both in-domain and OOD settings.

Abstract

Quadrupedal locomotion over complex terrain has been a long-standing research topic in robotics. While recent reinforcement learning-based locomotion methods improve generalizability and foot-placement precision, they rely on implicit inference of foot positions from joint angles, lacking the explicit precision and stability guarantees of optimization-based approaches. To address this, we introduce a foot position map integrated into the heightmap, and a dynamic locomotion-stability reward within an attention-based framework to achieve locomotion on complex terrain. We validate our method extensively on terrains seen during training as well as out-of-domain (OOD) terrains. Our results demonstrate that the proposed method enables precise and stable movement, resulting in improved locomotion success rates on both in-domain and OOD terrains.