Robot Routing Using Clustering-Based Parallel Genetic Algorithm with Migration

Abstract

Abstract-Wireless sensor network (WSN) technology is widely used in environment monitoring, health care, surveillance systems and unmanned space or planet exploration. To collect the exploration data, controllable mobility such as mobile robot called data mules can be used to help complete data gathering missions using the same sensors. This study focuses on data gathering by a mobile robot in a WSN, also referred to as a robot routing problem. It is formulated as a Traveling Salesman Problem with Neighborhoods (TSPN). Basing on the need of practical applications, a two-phase method, a clustering-based parallel genetic algorithm with migration (CBPGA), is proposed as a TSPN route computing and optimizing scheme that allows simultaneously searching of the order and locations of waypoints of clusters. The first phase of the method consists of a Euclidean-distance based clustering algorithm that assigns the nodes to clusters, and the second phase routes these clusters by planning the data collection points in the common intersection area of each cluster and the order to visit them using a genetic algorithm (GA). A travel cost reduction scheme by finding the closest waypoint in a cluster according to the visiting order is used to further shorten the path length. Simulation studies are conducted to evaluate the performance of CBPGA, by comparing it with other solutions to TSPN in WSN with identical or random sensing radii. To reveal the relative performance of each solution scheme, the effect of clustering, the role of GA and parallelism, and the integration of CBPGA in data-gathering route design are highlighted.