Applications of an adaptive hierarchical mobile robot navigation system

Abstract

A hierarchical robot navigation system has been developed and implemented on a range of real and simulated mobile robots for two independent research applications. One application is a study on emotion-modulated navigation. The other is an investigation of heterogeneous multi-robot navigation. The navigation framework shared by the various implementations is described. A multi-layered architecture is employed that incorporates reactive control, deliberative path planning and exploration capabilities. The reactive controller utilises a two-stage optimisation in directional and velocity space. Path planning employs a probabilistic A* algorithm applied to a dynamically-updated occupancy grid. Implementation differences between applications include the path planning cost function, and certain aspects of the reactive control system that are affected by different robot shapes and drive systems. Some key results from both applications are presented, including experiments performed in simulation and in the real world. 1