A Hierarchical Algorithm for Probabilistically Complete Path Planning in Multi-Floor Environments

Abstract

Navigation in multi-floor, multi-building environments is increasingly important in robotics. For wheeled robots, these environments can be conveniently modeled as a set of 2D maps, representing floors, connected by “wormholes”, which represent elevators and other between-floor connections. The full topological structure of the space can thus be described as a weighted graph. However, existing planning algorithms for multi-floor environments modeled in this way do not extend the guarantees on completeness and optimality provided by the underlying motion planning algorithms used within the 2D maps.

This work proposes a new algorithm, HRG*, for probabilistically complete and asymptotically optimal multi-floor path planning that carries these guarantees, together with a reference implementation whose performance is characterized in comparison to the native version.