A spline-based flexible method of virtual force design for dynamic motion planning of robots

Abstract

Motion planning for mobile robots is a problem on how a team of robots can achieve collective motion objectives in a shared working place while avoiding interference with one another. Its applications include logistics, military operations, and disaster rescue. As most applications take place in dynamic environments with frequently changing conditions, updating and maintaining route validity and optimality becomes a daunting task. There are some popular techniques for multi-robot motion planning, such as protocol-based methods, sampling-based techniques, genetic algorithms, spatiotemporal planning, potential field methods, virtual force approaches, graph-based techniques, etc. Each of these techniques has its own pros and cons. The virtual force approach is generally preferable for robots to perform team work in dynamic environments. This approach composes virtual attractive forces that drive the robots towards their targets and, at the same time, exert repulsion to steer the robots away from the obstructing robots. The design of virtual forces plays a pivotal role in improving robot motions to avoid robot collisions and enhance overall team performance. However, most traditional methods lack design flexibility to locally adjust the virtual forces according to the changing situations, hampering responsiveness of robot motions in dynamic environments and leading to possible robot collisions.