AI-DRIVEN SWARM INTELLIGENCE FOR COLLECTIVE ROBOTICS: A STUDY OF DECENTRALIZED CONTROL AND EMERGENT BEHAVIOR

Abstract

This research focused on the utilization of theoretical principles of Artificial Intelligence (AI) swarm intelligence that has a concentration on emergent behaviors and decentralization of control of robotic systems. The research concentrated on how robotic agents are able to self-organize, exhibit collaborative behaviors on a macro-level as a swarm, and maintain control on a micro-level through simple rules for a robot. The experiments used a varied composition of autonomous robots comprised of self-decision making, peer-to-peer (ai) and event driven architecture (sensors) to monitor the environment. The research concentrated on swarm behaviors like control of formations and exploration as well as simultaneous obstacle avoidance and dynamic workload (role) allocation within the swarm. The research documented the swarm behavior as highly decentralized self-adaptive to changing conditions and responsive to dynamic environments, demonstrating swarm level self-organization and scalability. The research documented range of control communication and the minimal number of agents and (low) swarm intelligence to maintain dynamic behavior of the swarm. The swarm demonstrated task completion on a (higher) level of efficiency (accuracy) and level of mastery (less variable). The research documented the potential of using swarm intelligence to create resilient systems of autonomous self-robotic entities. The research work documented emergent cooperative behaviors and proposed methodologies to decentralized control in robotic systems to be used in emergency response systems and automated production systems.