Swarm Control with RRT-APF Planning and FNN Task Allocation Tested on Mobile Differential Platform

This paper presents a novel method for centralized robotic swarm control that integrates path planning and task allocation subsystems. A swarm of agents is managed using various evaluation methods to assess performance. A feedforward neural network was developed to assign tasks to swarm agents in real time by predicting a suitability score. For centralized swarm planning, a hybrid algorithm combining Rapidly Exploring Random Tree (RRT) and Artificial Potential Field (APF) planners was implemented, incorporating a Multi-Agent Pathfinding (MAPF) solution to resolve simultaneous collisions at intersections. Additionally, experimental hardware using differential-drive, ArUco-tracked agents was developed to refine and demonstrate the proposed control solution. This paper specifically focuses on the swarm system design for applications in swarm reconfigurable manufacturing systems. Therefore, performance was evaluated on tasks that resemble such processes.