An enhanced frontier strategy with global search target-assignment approach for autonomous robotic area exploration

Abstract

Frontier strategy is an effective robotic area exploration mechanism that exploits the boundaries information between known area and unknown area to determine the next best target location for robots to explore autonomously. A typical frontier strategy employs a greedy-based local search approach to select a target location, also known as goal-assignment task, thus may slow down the exploration process. This paper presents a modified frontier strategy with a global search target-assignment paradigm. The proposed method optimises the target-assignment task by using genetic algorithm to provide a global search mechanism by carefully examining path distances between frontiers. A set of possible routes to visit all frontiers is generated heuristically by the genetic algorithm. After several generations, the first frontier of the shortest route is chosen as the next target location. The proposed enhanced frontier strategy outperforms the canonical frontier strategy in terms of the performance of area exploration by 31% to 50%.