A Hybrid Virtual Force Field Model For Autonomous Mobile Robot Navigation

ABSTRACT

This work improves upon the Potential Field Method (PFM) for the navigation of autonomous mobile robots, through the introduction of a new Hybrid Virtual Force Field (HVFF) concept. The HVFF concept integrates the virtual force field (VFF), which is based on the principles of artificial attractive and repulsive potentials, with the virtual obstacle concept (VOC) and the virtual goal concept (VGC). In particular, the specific challenges resolved by the HVFF include the local minimal problem posed by either lengthy or concave shaped obstacles as well as the potential field induced oscillatory motion of such a robot when maneuvering in the corridor between two narrowly spaced obstacles. These were all studied in a static obstacle domain. Furthermore, we extend our solution technique to the restructured obstacle environment all in a 2-D domain. In this context the general problem of dynamic obstacles in a completely unknown environment was examined in detail.

 Simulations of the motions were used to validate the efficacy of HVFF over existing algorithms for the dynamic and static obstacle architectures using Mobotsim (a customized software for robot animation).

Furthermore, we also confirm the feasibility of the new HVFF concept by demonstrating the performance of a prototype robotic vehicle designed, built and operated as an implementation platform for the HVFF algorithm.