Swarm Robot Foraging with Wireless Sensor Motes

Abstract

We investigate the use of wireless sensor motes as mobile deploy-able waypoints for swarm robot navigation in a foraging scenario. Each robot can deploy, retrieve, and optimize the location of the sensor motes. After deployment, the robots treat the sensor motes as nodes in a sparse graph, and store and retrieve multiple pheromones and flags locally in each of them. Pheromone information stored in the sensor motes allows the robots to build up gradients to dif-ferent targets of interest, and to determine which sensor motes are good candidates to optimize location, or to harvest for reuse else-where. Unlike many earlier pheromone-based foraging techniques, our method must deal with the physical reality of deploying and manipulating sensor motes, including a limited mote supply both on-board and in total, robustly dealing with occlusion and interference from other robots, and handling noise and robot or mote failure. We demonstrate the effectiveness of the technique both on differential drive robots of our own design, and in simulation, to examine its ability to robustly deal with various failure modes and changes in environment.