Echo in the Steps: Learning Perceptive Humanoid Parkour with Gated Memory

2026-09-23

AI Summary

This study addresses the challenge of agile parkour for humanoid robots in highly discontinuous environments, which requires precise foothold selection under sparse conditions and rapid alternating gaits. To this end, it proposes an end-to-end control framework relying solely on onboard depth perception. The framework introduces a saliency-guided temporal-aware module to extract critical features and incorporates a novel gated memory mechanism to aggregate multi-frame depth information. Furthermore, an alternating loss function is constructed to enforce symmetric regularization, ensuring gait consistency during rapid transitions. Both simulation and real-world experiments demonstrate that the proposed approach significantly improves task success rates and foothold placement accuracy over challenging terrains, validating the robustness and practical utility of the system.

Abstract

While recent advances in perceptive locomotion have enabled humanoid robots to traverse structured terrains, agile parkour in highly discontinuous environments remains an open challenge. In particular, crossing sparse footholds and narrow support regions requires precise foothold selection, effective use of visual observations, and consistent alternating foot placement during fast transitions. In this paper, we present a perceptive humanoid parkour framework that enables stable traversal across terrains with limited foothold availability using only onboard depth observations. The framework features a saliency-guided temporal perception module that combines a saliency prior with gated memory. It retains informative depth features across frames, enabling reliable foot placement from partial observations. By introducing an alternation loss, our symmetry regularization encourages alternating gait patterns and improves traversal robustness. Extensive experiments show that our method significantly improves success rate and foothold accuracy on challenging terrains in both simulation and the real world.

Problem

• humanoid parkour
• perceptive locomotion
• foothold selection
• discontinuous environments

Innovation

• Humanoid Parkour
• Gated Memory
• Saliency-guided Perception
• Alternation Loss
• Perceptive Locomotion