Moving Together: Coordination and Avoidance for Human-Aware Robots and Drone Swarms

Autonomous agents, whether on the ground or in the air, must navigate shared spaces, avoiding obstacles or coordinating within a group. Effective navigation in dynamic environments requires adaptive methods that ensure safety while achieving the agents' goals. In this talk I will present two of our latest results in multi-agent navigation: AVOCADO and Gen-Swarms. AVOCADO is a new reactive navigation algorithm that uses opinion dynamics to enable robots to navigate seamlessly through human crowds, adapting in real time to the cooperation levels of surrounding agents while ensuring safe and efficient collision avoidance. Gen-Swarms, on the other hand, uses generative AI to help non-expert users easily define desired swarm configurations, producing smooth trajectories and accounting for potential collisions through a reactive navigation algorithm. The talk will show how these approaches offer the ability to react intelligently to dynamic surroundings and the power to generate structured, large-scale behaviors from high-level user inputs.