Brownian Motion as Exploration Strategy for Autonomous Swarm Robots

The use of swarms robots to search in collapsed environments, today becomes in a real alternative solution to support rescue efforts in landslide and other disasters. However, these environments have special features (unknown and dynamic) under which rapid responses are required (little processing time) with a major limitation of sensors (problems for the use of cameras and GPS's for example) and communication. In this paper, we analyze an alternative of navigation for a swarm robots with very limited capabilities (very limited processing power, communication and sensing). Our minimalist approach seeks to solve the problem without requiring system identification, geometric map building, localization, or state estimation. Instead, we propose a strategy based on Brownian motion, in which each robot is modeled as a particle whose motion is influenced by landmarks installed in the environment. The degree of influence on the robot corresponds to the design of the navigation route. Under this scheme, the robots perform minimal processing, and the parallel navigation increases confidence in the search process. The proposed navigation scheme is analyzed and then evaluated by simulation.