Bipedal humanoids have a unique advantage over fixed-base robots: they can change their effective footprint by rotating their body. A new paper on arXiv proposes a posture-aware navigation framework that exploits this ability, using what the authors call adaptive traversal tubes to plan paths through cluttered or socially dynamic environments.
The core idea is that a robot's orientation is not just a byproduct of movement but a planning variable. By adjusting body yaw, the robot can reduce its footprint to slip through narrow passages, or present a more accommodating profile when yielding to people. This makes navigation more flexible than approaches that treat the robot as a simple point or rigid box.
The framework is designed for bipedal humanoids specifically, and the abstract highlights two benefits: narrow-passage traversal and socially aware yielding. The paper is available as arXiv:2610.27688. As this is a single preprint abstract, no independent verification or comparison with other methods is included.