Capability-Conditioned Graph Control Barrier Functions for Zero-Shot Safe Navigation of Heterogeneous Mobile Robot Teams

1. 05:00
   arXiv Robotics — research abstractsSignal

   Capability-Conditioned Graph Control Barrier Functions for Zero-Shot Safe Navigation of Heterogeneous Mobile Robot Teams

   This research proposes the Capability-Conditioned Heterogeneous Graph Control Barrier Function (CC-HGCBF) and Radial Capability Field (RCF) to enable zero-shot safe navigation for heterogeneous mobile robot teams. The method defines a shared second-order reference-point operational space that provides a common geometric representation of safety across diverse robots while retaining model-specific dynamics. By conditioning the shared graph barrier and decentralized policy on the RCF, the system adapts to state-dependent motion capabilities and control limits without relying on predefined robot identities. Experiments demonstrate scalable safe coordination and zero-shot transfer to both unseen robot models and capability variations withheld during training, with physical-robot experiments confirming deployment to unseen kinematic models without retraining.

Current attention 9·Peak within the comparable range 10（2026-11-02 06:00 UTC）·Change within the comparable range over 24 hours –

The trend compares participants observed continuously; its scope may be narrower than current attention totals. Hover or click to explore hourly observations, or use the left and right arrow keys.