Motion planning and Collision Avoidance using Non-Gradient Vector Fields. Technical Report

TL;DR

This paper introduces a novel vector field-based feedback method for motion planning and collision avoidance in unicycle robots, extending to multi-robot systems with local information, ensuring safety and convergence without local minima issues.

Contribution

The paper presents a new analytic vector field approach for motion planning that avoids local minima and extends to distributed multi-robot coordination with safety guarantees.

Findings

Effective in static obstacle environments

Ensures collision-free multi-robot coordination

Demonstrated via simulation results

Abstract

This paper presents a novel feedback method on the motion planning for unicycle robots in environments with static obstacles, along with an extension to the distributed planning and coordination in multi-robot systems. The method employs a family of 2-dimensional analytic vector fields, whose integral curves exhibit various patterns depending on the value of a parameter lambda. More specifically, for an a priori known value of lambda, the vector field has a unique singular point of dipole type and can be used to steer the unicycle to a goal configuration. Furthermore, for the unique value of lambda that the vector field has a continuum of singular points, the integral curves are used to define flows around obstacles. An almost global feedback motion plan can then be constructed by suitably blending attractive and repulsive vector fields in a static obstacle environment. The method does not suffer from the appearance of sinks (stable nodes) away from goal point. Compared to other similar methods which are free of local minima, the proposed approach does not require any parameter tuning to render the desired convergence properties. The paper also addresses the extension of the method to the distributed coordination and control of multiple robots, where each robot needs to navigate to a goal configuration while avoiding collisions with the remaining robots, and while using local information only. More specifically, based on the results which apply to the single-robot case, a motion coordination protocol is presented which guarantees the safety of the multi-robot system and the almost global convergence of the robots to their goal configurations. The efficacy of the proposed methodology is demonstrated via simulation results in static and dynamic environments.