Swarm robotics obstacle avoidance: A progressive minimal criteria novelty search-based approach

Abstract

Swarm robots are required to explore and search large areas. In order to cover largest possible area while keeping communications, robots try to maintain hexagonal formation while moving. Obstacle avoidance is an extremely important task for swarm robotics as it saves robots from hitting objects and being damaged. This paper introduces novelty search evolutionary algorithm to swarm robots multi-objective obstacle avoidance problem in order to overcome deception and reach better solutions. This work could teach robots how to move in different environments with 2.5% obstacles coverage while keeping their connectivity more than 82%. Percentage of robots reached the goal was more than 97% in 70% of the environments and more than 90% in the rest of the environments.