Multi-Agent Ergodic Coverage with Obstacle Avoidance

Autonomous exploration and search have important applications in robotics. One interesting application is cooperative control of mobile robotic/sensor networks to achieve uniform coverage of a domain. Ergodic coverage is one solution for this problem in which control laws for the agents are derived so that the agents uniformly cover a target area while maintaining coordination with each other. Prior approaches have assumed the target regions contain no obstacles. In this work, we tackle the problem of static and dynamic obstacle avoidance while maintaining an ergodic coverage goal. We pursue a vector-field-based obstacle avoidance approach and define control laws for idealized kinematic and dynamic systems that avoid static and dynamic obstacles while maintaining ergodicity. We demonstrate this obstacle avoidance methodology via numerical simulation and show how ergodicity is maintained. Keywords—Multi-agent planning, centralized robot control, ergodic theory, uniform coverage, obstacle avoidance.