Potential field based approach for multi-threat containment with cooperative robots

This work investigates a multithreat containment problem where a team of autonomous robots cooperate to contain occurring threats. The robots aim to form circles around multiple threats that appear at random times and expire after a certain time. Containing single static threats has been studied, yet extending it to the dynamic regime with threat lifetime overlapping presents additional complex challenges. An algorithm named Multiple Threat Containment Algorithm (MUTCA) is proposed and analyzed in this work. MUTCA is designed for simplicity and based on Quadratic Artificial Potential Functions (QAPF) and the use of random walks. Multiple QAPFs are carefully designed to allow robots to move toward the threats, to avoid collisions, and to surround the threats evenly. This work investigates the capabilities and limitations of autonomous robots utilizing MUTCA via simulation using MAHESHDAS - a simulator developed to simulate autonomous robots. Application specific success rates are examined for various scenarios and robot sensor capabilities. MUTCA has been found to work well with sufficient robots that have moderate capabilities, but is sensitive to application specific parameters.