Bioinspired Obstacle Avoidance Algorithms for Robot Swarms

Recent work in socio-biological sciences have introduced simple heuristics that accurately explain the behavior of pedestrians navigating in an environment while avoiding mutual collisions. We have adapted and implemented such heuristics for distributed obstacle avoidance in robot swarms, with the goal of obtaining human-like navigation behaviors which would be perceived as friendly by humans sharing the same spaces. In this context, we study the effects of using different sensing modalities and robot types, and introduce robot’s emotional states, which allows us to modulate system’s group behavior. Experimental results are provided for both real and simulated robots. The extensive quantitative simulations show the macroscopic behavior of the system in various scenarios, where we observe emergent collective behaviors – some of which are similar to those observed in human crowds.