🤖 AI Summary
To address the challenge of precise foot placement for quadrupedal robots navigating sparse, high-risk terrain, this paper proposes an end-to-end reinforcement learning framework relying solely on proprioceptive sensing and forward-facing monocular depth images. The core innovation is a differentiable local terrain reconstruction module that transforms raw depth maps into structured, noise-robust local heightmaps—serving as an intermediate visual representation bridging feature extraction and locomotion control. A lightweight multimodal policy network is then constructed by fusing proprioceptive inputs with the reconstructed heightmap. Evaluated on a low-cost quadruped platform, the method enables real-time, adaptive traversal across diverse, unstructured real-world terrains. In sparse, randomly sampled foothold scenarios, it achieves a 42% higher success rate compared to vision-only or proprioception-only baselines.
📝 Abstract
Traversing risky terrains with sparse footholds presents significant challenges for legged robots, requiring precise foot placement in safe areas. To acquire comprehensive exteroceptive information, prior studies have employed motion capture systems or mapping techniques to generate heightmap for locomotion policy. However, these approaches require specialized pipelines and often introduce additional noise. While depth images from egocentric vision systems are cost-effective, their limited field of view and sparse information hinder the integration of terrain structure details into implicit features, which are essential for generating precise actions. In this paper, we demonstrate that end-to-end reinforcement learning relying solely on proprioception and depth images is capable of traversing risky terrains with high sparsity and randomness. Our method introduces local terrain reconstruction, leveraging the benefits of clear features and sufficient information from the heightmap, which serves as an intermediary for visual feature extraction and motion generation. This allows the policy to effectively represent and memorize critical terrain information. We deploy the proposed framework on a low-cost quadrupedal robot, achieving agile and adaptive locomotion across various challenging terrains and showcasing outstanding performance in real-world scenarios. Video at: youtu.be/Rj9v5EZsn-M.