Machine Learning Enabled Colour Light Communication For Robot Swarm Coordination

Abstract

Machine Learning Enabled Colour Light Communication for Robot Swarm Coordination newline newlineAbhishek Kaushal newline newlineABSTRACT newline newlineSwarm robotics, inspired by the collective behaviour observed in natural swarms, offers promising solutions to complex tasks through the coordination of multiple robots, or agents. Effective communication among these agents is essential for achieving desired swarm behaviour and overall system performance. This thesis addresses the challenge of enhancing inter-robot communication within swarm robotics systems through the novel integration of machine learning techniques with colour light-based communication protocols. newline newlineThe research begins by recognizing the inherent complexity in current swarm robotics systems, particularly in their control and communication mechanisms. Building upon this understanding, the thesis proposes a new approach leveraging visible light-based communication techniques, which offer advantages in scalability and simplicity. The proposed system facilitates seamless interaction among individual robotic agents, fostering emergent swarm behaviours crucial for task accomplishment. newline newlineCentral to the thesis is the development of a machine learning-enabled colour recognition system tailored for miniature mobile swarm robots. Supervised machine learning models, including XGBoost, are evaluated for their efficacy in accurately recognizing and interpreting colour-encoded signals. Through rigorous experimentation, XGBoost emerges as the top performer, achieving a classification accuracy of 96.66% while maintaining efficient execution times and a compact memory footprint suitable for embedded microcontrollers. newline newlineFurthermore, the thesis explores various swarming behaviours orchestrated through the enhanced communication capabilities facilitated by machine learning. These behaviours serve as proof of concept, demonstrating the feasibility and effectiveness of the proposed approach in enabling coordinated actions among robotic swarms. By harnessing the synergy between machine learning and colour light c