Hetergenous Getting out Maneuver
Diverse methods for strategic evasion and escape, addressing varying challenging environments.
During my research project on “Multi-Agent Planning” under the mentorship of Prof. Matthew from January 2023 to June 2023, the central focus was on innovating an efficient egress maneuver for autonomous agents. The primary objective was to craft a maneuver capable of swiftly evacuating an agent from a designated area within a concise timeframe of no more than 4 seconds. This innovation held critical importance in minimizing disruptions to overall task completion speed and enhancing user operability by significantly reducing the time agents spent in egress mode.
To assess the maneuver’s efficacy, a series of experiments were conducted in environments laden with obstacles, including open spaces as narrow as 0.5 meters. These challenging scenarios aimed to replicate real-world situations where an agile and adaptable egress strategy is indispensable for multi-agent systems. The results unequivocally showcased the maneuver’s success in safely guiding agents through these intricate environments. Furthermore, the study was extended to encompass various agent types, ranging from RC cars to Boston Dynamics Spot robots, ensuring the maneuver’s broad applicability across a diverse spectrum of autonomous agents. This project not only contributed significantly to the field of multi-agent planning but also underscored the pivotal importance of crafting agile and adaptable egress strategies for autonomous systems operating in diverse and dynamic environments.
