A Modified Dijkstra Algorithm for ROS Based Autonomous Mobile Robots

Abstract

Autonomous Mobile Robots (AMRs) are frequently used in many fields of technology. In this study, an AMR was designed to execute different path planning algorithms. Firstly, working principle, system architecture and motion planning of AMR are presented. Then, a map for the current environment is produced by a Robot Operating System (ROS) powered AMR which was designed for this study. The AMR locates itself on the produced map with the aid of an integrated Light Detection and Ranging sensor (LIDAR). The locomotion of AMR to a user-defined target on the produced map is performed by an optimal path based on AMR's own navigation plan. Two different path planning algorithms, which are Dijkstra’s algorithm and a modified version of Dijkstra’s algorithm, are executed on a cost-effective AMR platform, which has the capability of Simultaneous Localization and Mapping (SLAM). The reason why Dijkstra algorithm is handled in this study rather than A*, D* and RRT algorithms is that this algorithm is a basic and widely used algorithm. Dijkstra’s algorithm is modified, and pros and cons of the modified algorithm are analysed compared to Dijkstra algorithm. The proposed algorithm and navigation of AMR are tested both in real time in real world and as a simulation in Gazebo. Two algorithms were compared according to the results obtained from the robot locomotion both in real application and simulation environment. It is observed that the modified version of the Dijkstra’s algorithm comparatively yielded a bit more satisfactory results in the aspect of path planning.