# Global-Local Attention Decomposition for Terrain Encoding in Humanoid Perceptive Locomotion

arXiv: https://arxiv.org/abs/2606.00637 · HTML: https://mlsanity.com/paper/2606.00637/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in RL for robot locomotion and manipulation, one of 15 directions in Deep reinforcement learning methods.
- Builds on 22 of its 27 references that are in the map.
- First posted on arXiv 30 May 2026.

## Placed in

- Deep reinforcement learning methods / [RL for robot locomotion and manipulation](https://mlsanity.com/deeprl/rl-for-robot-locomotion-and-manipulation/)

## Builds on (most cited first, 22 of 22)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learning](https://mlsanity.com/paper/2109.11978/) (2021)
- [Real-World Humanoid Locomotion with Reinforcement Learning](https://mlsanity.com/paper/2303.03381/) (2023)
- [Learning Vision-Guided Quadrupedal Locomotion End-to-End with Cross-Modal Transformers](https://mlsanity.com/paper/2107.03996/) (2021)
- [FAST-LIO: A Fast, Robust LiDAR-inertial Odometry Package by Tightly-Coupled Iterated Kalman Filter](https://mlsanity.com/paper/2010.08196/) (2020)
- [Learning Humanoid Locomotion with Perceptive Internal Model](https://mlsanity.com/paper/2411.14386/) (2024)
- [BeamDojo: Learning Agile Humanoid Locomotion on Sparse Footholds](https://mlsanity.com/paper/2502.10363/) (2025)
- [Elevation Mapping for Locomotion and Navigation using GPU](https://mlsanity.com/paper/2204.12876/) (2022)
- [Learning Vision-Based Bipedal Locomotion for Challenging Terrain](https://mlsanity.com/paper/2309.14594/) (2023)
- [Attention-Based Map Encoding for Learning Generalized Legged Locomotion](https://mlsanity.com/paper/2506.09588/) (2025)
- [Learning Humanoid Locomotion over Challenging Terrain](https://mlsanity.com/paper/2410.03654/) (2024)
- [PIE: Parkour with Implicit-Explicit Learning Framework for Legged Robots](https://mlsanity.com/paper/2408.13740/) (2024)
- [Learning Agile Locomotion on Risky Terrains](https://mlsanity.com/paper/2311.10484/) (2023)
- [TAMOLS: Terrain-Aware Motion Optimization for Legged Systems](https://mlsanity.com/paper/2206.14049/) (2022)
- [Hiking in the Wild: A Scalable Perceptive Parkour Framework for Humanoids](https://mlsanity.com/paper/2601.07718/) (2026)
- [Learning Perceptive Humanoid Locomotion over Challenging Terrain](https://mlsanity.com/paper/2503.00692/) (2025)
- [DPL: Depth-only Perceptive Humanoid Locomotion via Realistic Depth Synthesis and Cross-Attention Terrain Reconstruction](https://mlsanity.com/paper/2510.07152/) (2025)
- [RPL: Learning Robust Humanoid Perceptive Locomotion on Challenging Terrains](https://mlsanity.com/paper/2602.03002/) (2026)
- [Walking with Terrain Reconstruction: Learning to Traverse Risky Sparse Footholds](https://mlsanity.com/paper/2409.15692/) (2024)
- [Gait-Adaptive Perceptive Humanoid Locomotion with Real-Time Under-Base Terrain Reconstruction](https://mlsanity.com/paper/2512.07464/) (2025)
- [Agile and Generalized Legged Locomotion via Attention-Based Neural Map Encoding](https://mlsanity.com/paper/2601.08485/) (2026)
- [No More Blind Spots: Learning Vision-Based Omnidirectional Bipedal Locomotion for Challenging Terrain](https://mlsanity.com/paper/2508.11929/) (2025)

## Built on by (most cited first, 1 of 1)

- [FootQuery: Future-Touchdown-Guided Retrieval from Depth History for Perceptive Humanoid Locomotion](https://mlsanity.com/paper/2609.21447/) (2026)

## Abstract (arXiv)

Although reinforcement learning has significantly advanced humanoid locomotion, perceptive policies still struggle on sparse-foothold terrain and constrained environments. Success in these scenarios requires both broad terrain awareness and precise foothold selection, two perceptual roles that conventional encoders often entangle. To address this challenge, we propose Global-Local Attention Decomposition (GLAD) for terrain encoding in humanoid locomotion. Realized by a coarse-to-fine encoder over a robot-centric elevation map, GLAD explicitly separates these objectives: a global attention branch uses attention pooling to summarize the surrounding terrain context, while a local attention branch sparsifies the local features by terrain saliency and applies state-conditioned attention to encode precise foothold-relevant geometry. This explicit attention decomposition prevents the dilution of fine-grained spatial cues while reducing training overhead. Experiments demonstrate that GLAD enables reliable locomotion over challenging gaps, stepping stones, and stairs. Furthermore, the learned policy exhibits emergent terrain-responsive behaviors, autonomously following narrow paths and avoiding obstacles under forward-velocity commands alone, without explicit navigation planners. In real-world deployment on a Unitree G1 humanoid robot using onboard LiDAR, the proposed method achieves robust zero-shot sim-to-real transfer across diverse sparse-foothold and obstacle-rich domains.