aiqbench analysis
This research introduces a lightweight preprocessing method for humanoid locomotion that anticipates foot placement to reduce redundant visual input, significantly accelerating training convergence. By sampling elevation maps only around predicted footholds, the approach outperforms full-map observation policies in sparse terrain scenarios while matching expensive attention-based encoders at a fraction of the cost. The technique demonstrates strong zero-shot transfer capabilities on diverse obstacles and environments for Unitree G1 humanoids.
Watch Your Step: Anticipated Perception for Humanoid Locomotion
arXiv:2610.29612v1 Announce Type: cross Abstract: Perceptive locomotion policies typically observe a dense robot-centric elevation map, although only the terrain beneath and around the next footsteps informs the control decision. Learned encoders compress the whole visual input, leaving the policy to work out which inputs matter while learning the baseline locomotion task. Learning both at once slows training and, without computationally expensive attention mechanisms, makes generalization to new terrain harder. We instead remove redundant inputs from the elevation map before the policy sees it. A lightweight foothold prediction model (FPM) predicts future footholds from proprioception and the velocity command alone, and we sample the elevation map only in small patches around these predicted footholds. We train the prediction model alongside the locomotion policy, periodically retraining it on new policy rollouts as the robot's gait matures. This allows the perception to anticipate where the policy will step, while the policy learns to leverage terrain information around those steps. With the proposed lightweight preprocessing, the policy reaches a given terrain difficulty in roughly a third of the training iterations required with the complete map. It outperforms policies that observe the complete map where footholds are sparse, including in narrow walled corridors, and on unseen terrain, and it matches an attention-based encoder trained at several times the cost. The policy transfers zero-shot to a Unitree G1 humanoid traversing various obstacles, stairs, gaps, and step-downs beyond the training distribution in both indoor and outdoor environments.
Abstract:Perceptive locomotion policies typically observe a dense robot-centric elevation map, although only the terrain beneath and around the next footsteps informs the control decision. Learned encoders compress the whole visual input, leaving the policy to work out which inputs matter while learning the baseline locomotion task. Learning both at once slows training and, without computationally expensive attention mechanisms, makes generalization to new terrain harder. We instead remove redundant inputs from the elevation map before the policy sees it. A lightweight foothold prediction model (FPM) predicts future footholds from proprioception and the velocity command alone, and we sample the elevation map only in small patches around these predicted footholds. We train the prediction model alongside the locomotion policy, periodically retraining it on new policy rollouts as the robot's gait matures. This allows the perception to anticipate where the policy will step, while the policy learns to leverage terrain information around those steps. With the proposed lightweight preprocessing, the policy reaches a given terrain difficulty in roughly a third of the training iterations required with the complete map. It outperforms policies that observe the complete map where footholds are sparse, including in narrow walled corridors, and on unseen terrain, and it matches an attention-based encoder trained at several times the cost. The policy transfers zero-shot to a Unitree G1 humanoid traversing various obstacles, stairs, gaps, and step-downs beyond the training distribution in both indoor and outdoor environments.
8 pages, 5 figures, 3 tables. Project page: this https URL
Robotics (cs.RO); Machine Learning (cs.LG)
arXiv:2610.29612 [cs.RO]
(or arXiv:2610.29612v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.29612 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 342 words
Related coverage
FootQuery: Future-Touchdown-Guided Retrieval from Depth History for Perceptive Humanoid Locomotion
arXiv · Sep 21, 2026
Learning from Next-Frame Prediction: Towards Unified and Scalable Visual Representation Learning
arXiv · Oct 23, 2026
OmniRobotHome: A Multi-Camera Home Platform for Real-Time Human-Robot Interaction
arXiv · Oct 14, 2026
Anticipatory Risk-Guided Reinforcement Learning for Safe Flight Through Dynamic Clutter
arXiv · Jul 28, 2026