Adaptive Traversal Tubes for Posture-Aware Navigation of Bipedal Humanoids

Zijian Ge

cs.ROOct 20, 2026

Adaptive Traversal Tubes for Posture-Aware Navigation of Bipedal Humanoids

Authors: Zijian Ge

Abstract

Bipedal humanoids can change their effective footprint by adjusting body yaw, enabling both narrow-passage traversal and socially aware yielding. This paper presents a posture-aware navigation framework centered on Adaptive Traversal Tubes (ATT), Interval Velocity--Heading (IVH) control, and a sidestep-enhanced AMP locomotion prior. ATT represents a behavior-feasible corridor width field along a reference path, unifying geometric clearance and social comfort into a continuous signal. IVH maps this width to a continuous frontal-to-sidestep posture schedule and omnidirectional velocity commands. A velocity-conditioned AMP policy, trained with GVHMR-retargeted frontal/sidestep motions and coupled $(v_x,v_y)$ tracking rewards, is designed to improve transition stability over a generic locomotion controller while remaining interchangeable with the default onboard policy. This preprint presents the full method; a planned evaluation protocol on a Unitree G1 is outlined, with quantitative results to follow. Demo videos of MuJoCo simulation and real-robot deployment are available at: github.com/lborogzj997/Humanoid-Navigation-in-Crowds-and-Narrow-Spaces.

Figures & tables

Fig. 1: Overview of the proposed idea. Top: ATT width field (green $\,\to\,$ red as the tube narrows) for a geometric squeeze-through (left) and social comfort compression (right). Bottom: corresponding real-robot behaviors on a Unitree G1, where IVH schedules a frontal-to-sidestep posture response.

Fig. 2: System overview. Mid-360 LiDAR localization supports global A* and local Hybrid A* planning. ATT builds a behavior-feasible width field $w(s)$ along the reference path from geometric clearance and social comfort; IVH maps $w(s)$ to a frontal–sidestep heading schedule and body-frame $(v_{x},v_{y},\omega_{z})$ for a sidestep-enhanced AMP locomotion policy. Offline, GVHMR+GMR provide frontal/sidestep reference motions for AMP training; a generic onboard controller remains a compatible fallback.

Fig. 3: Qualitative illustration of the intended behaviors on a Unitree G1 (not a quantitative benchmark). (a) Socially aware pedestrian encounter: comfort-space compression narrows the tube and IVH schedules a frontal-to-sidestep yield. (b) Narrow-corridor traversal: geometric clearance shrinks $w(s)$ at the gap, triggering a sidestep squeeze-through and recovery to frontal gait. Controlled comparisons against a frontal-only baseline are left to the evaluation protocol in Section IV .

Robot Navigation

Robot navigation studies how robots perceive, localize, plan, and move safely toward goals in static or changing environments.

See all 554 papers

SSR: Scaling Surefooted and Symmetric Humanoid Traversal to the Open World
May 29, 2026Ruiqi Yu, Yiwen Wang, Yuan Hao +2Humanoid Robot Locomotion Terrain-Aware Robot Locomotion
Zhejiang University
GuideWalk: Learning Unified Autonomous Navigation and Locomotion for Humanoid Robots across Versatile Terrains
Jun 9, 2026Haoxuan Han, Chen Chen, Linao Gong +6Humanoid Robot Locomotion Robot Navigation
Harbin Institute of Technology · Leju Robotics
ADAPT: Adaptive Dual-projection Architecture for Perceptive Traversal
Mar 17, 2026Shuo Shao, Tianchen Huang, Wei Gao +1Humanoid Robot Locomotion Terrain-Aware Robot Locomotion
Department of Automation, University of Science and Technology of China, Anhui 230026, China. · Institute of Humanoid Robots, Department of Precision Machinery and Precision Instrumentation, University of Science and Technology of China, Hefei, Anhui 230026, China.

May 29, 2026J/K move · Enter open · S save

SSR: Scaling Surefooted and Symmetric Humanoid Traversal to the Open World

Ruiqi Yu, Yiwen Wang, Yuan Hao, Jun WU+1

Zhejiang University

Humanoid Robot Locomotion Terrain-Aware Robot Locomotion Robot Locomotion Equivariant Representation Learning Vision-Based Robot Control

Open paper AskPDF arXiv

Extending humanoid traversal to the open world is key to practical deployment in human environments, but remains challenging. The robot must use vision to ensure safe and reliable foot placement on heterogeneous terrain under highly dynamic motion, while producing coordinated, natural whole-body behaviors. We propose SSR, an efficient end-to-end framework for egocentric vision-based humanoid traversal that jointly learns these capabilities. SSR introduces imagined foothold guidance, which learns to model forthcoming swing-foot contacts and evaluates their support to guide pre-touchdown swings toward stable regions, reducing edge slips. It further employs equivariant latent-space symmetry augmentation to efficiently induce bilateral coordination under high-dimensional visual observations, and uses terrain-specific multi-discriminator motion priors to encourage human-like behavior across scenes. Extensive experiments show that SSR achieves safe, stable, and high-quality locomotion on diverse real-world terrains, including stairs with varied structures and extreme challenges such as wide gaps and high platforms, while enabling reliable long-horizon traversal in open outdoor environments.

Abstract

Robot Navigation

Explore similar work

SSR: Scaling Surefooted and Symmetric Humanoid Traversal to the Open World

GuideWalk: Learning Unified Autonomous Navigation and Locomotion for Humanoid Robots across Versatile Terrains

ADAPT: Adaptive Dual-projection Architecture for Perceptive Traversal