Development of an Artificial Potential Field-Based Navigation System for a Swarm Robot Project

Abstract

This paper presents a reactive potential field method for coordinated navigation of leader-follower mobile robots in dynamic environments. The leader navigates toward goals using attractive and repulsive forces, while the follower maintains formation by treating the leader as a dynamic target with avoidance constraints. The method integrates vortex forces for local minimum escape and adaptive velocity control. Vision-based localization using ArUco markers provides real-time pose estimation with homography-based coordinate transformation. Experimental validation was conducted on ESP8266-based differential-drive robots across 15 trials in three scenarios within a 200×150 cm arena. Results demonstrate 93.3% leader goal-reaching success with mean path efficiency of 0.86. The system achieved 80% collision-free navigation with follower formation error of 18 cm from the target 25 cm following distance.