Potential Gap: A Gap-Informed Reactive Policy for Safe Hierarchical Navigation

This letter considers the integration of gap-based local navigation methods with artificial potential field (APF) methods to derive a local planning module, called potential gap, for hierarchical navigation systems. Central to the construction of the local planner is the use of sensory-derived local free-space models that detect gaps and use them for the synthesis of the APF. Trajectories derived from the APF are provably collision-free for idealized robot models. The provable property is lost when applied to more realistic models. A set of algorithm modifications correct for these errors and enhance robustness to non-ideal models, in particular a nonholonomic robot model. Integration of the potential gap local planner into a hierarchical navigation system provides the local goals and trajectories needed for collision-free navigation through unknown environments. Monte Carlo experiments in benchmark worlds confirm the asserted safety and robustness properties.

Automated summary

This study proposes a local planning module called "potential gap" that integrates gap-based local navigation methods with artificial potential field (APF) methods for hierarchical navigation systems. By using sensory-derived local free-space models to detect gaps and synthesize the APF, collision-free trajectories can be achieved. Algorithm modifications are introduced to correct errors and enhance robustness for non-ideal models, particularly nonholonomic robot models. Integration of the potential gap local planner into hierarchical navigation systems provides local goals and trajectories for collision-free navigation through unknown environments, as confirmed by Monte Carlo experiments in benchmark worlds.