Path planning algorithm for a two-wheel differential mobile robot in robot operating system

This thesis focuses on the algorithm used for navigation between two different locations by avoiding obstacles. The algorithm is developed in a way that results in an efficient path planner between two given locations. Algorithms such as the potential field approach and Voronoi diagram are used as references in developing the suggested algorithm by a hybrid approach to developing the performance of the proposed algorithm. Python is used to write a new path planner function for the algorithm developed under this research work. Navigation is achieved by integrating the proposed path planner to the open-source middleware called Robot Operating System (ROS). A custom robot is defined in the ROS ecosystem using the Unified Robot Description Format (URDF). Several launch files are created to launch the robot in the desired environment and launch the path planning algorithm in Gazebo. Localization is achieved using the open-source gmapping plugin in ROS. An occupancy grip map of the domain is obtained using RVIZ software with custom-built parameters of the laser attached to the robot. The proposed algorithm is then integrated into ROS by creating several nodes involving publisher-subscriber and server-client relations to make it available for the robot to receive information from its sensors and send velocity commands to navigate effectively. A prior map is provided for the path planner node in ROS to enable the robot to plan its desired line of the path towards the goal location. The desired line of the track is maintained in conjunction with real-time sensor readings. The frequencies of path planners are determined such that it enables a successful obstacle avoidance solution. The performance of the proposed algorithm concerning path length and execution time is weighed with some of the well-known algorithms. The kinematic and dynamic constraints of the differential mobile robot are not considered in this approach