The Role of Environmental and Controller Complexity in the Distributed Optimization of Multi-Robot Obstacle Avoidance

摘要: The ability to move in complex environments is a fundamental requirement for robots be part of our daily lives. Increasing the controller complexity may desirable choice order obtain an improved performance. However, these two aspects pose considerable challenge on optimization robotic controllers. In this paper, we study trade-offs between reactive controllers and environment multi-robot obstacle avoidance resource-constrained platforms. carried out simulation using distributed, noise-resistant implementation Particle Swarm Optimization, resulting are evaluated both with real robots. We show that simple environment, linear only parameters perform similarly more non-linear up twenty parameters, even though latter ones require evaluation time learned. complicated there increase performance when can differentiate front backwards sensors, but increasing further number sensors adding activation functions provide no benefit. environments, augmenting control laws memory capabilities causes highest also measurements noisier, optimal parameter region smaller, iterations required process converge.