Resilient Multiagent Robotic Systems

Several mission applications that involve the deployment of groups of autonomous vehicles demand decentralized swarming capabilities and require advanced and novel technologies to increase overall mission performance, particularly if they are operating under complex and dynamically changing environments. These advanced autonomous systems based on single-agent or multi-agent cooperative networks require adequate intelligent systems to increase mission safety and optimize performance within complex, unstructured and dynamic operating environments. The purpose of this project is to develop intelligent algorithms to improve resilience of multi-agent systems (MAS). A general architecture inspired by the functioning of biological mechanisms is designed to increase autonomy in a swarm of unmanned aerial vehicles (UAVs) through self-learning, self-organizing and optimal flight trajectories. We demonstrate and evaluate on-board intelligent techniques at different levels by combining self-detection, self-diagnosis, self-recovery and self-organization as a means to increase the safety of swarm missions, optimize endurance, and maintain performance within hazardous operating environments with minimal human intervention.