Development and implementation of an obstacle avoidance algorithm for an Autonomous Mobile Robot

The ongoing Industrial Revolution necessitates a high level of automation in order to be flexible and adaptive enough to comply with the demands of the market. Autonomous and collaborative robots will play an ever-greater role in this context. Therefore, the FIXIT project is proposed by CIM 4.0 with the goal of providing interactive support for the human operator within an industrial or logistic environment to meet the Industry 4.0 requirements. The objective of this thesis is to develop and implement path planning algorithm with obstacle avoidance technology to be executed on the AMR of FIXIT. Firstly, the state of the art of path planning methods is introduced to find the most suitable one for our project. Secondly, the URDF model is built, based on which the autonomous navigation is simulated by Gazebo and the results are visualized by RViz. Lastly, a real robot with a three-layer mechanical system: chassis, control, and application layers; and a two-layer control system: a higher-level distributed computer system and a lower-level motion control system, is designed to perform the path planning algorithms. To realize the ability of avoiding moving obstacles, D* Lite with high adaptability to dynamic environments, is chosen as the GPP algorithm, while a DAPF is proposed as the LPP algorithm, which takes into consideration the moving obstacle detected by OpenCV tools such as background subtraction and blob detection, and tracked by the Kalmann filter estimating its position and velocity. The feasibility of the algorithm is verified by MATLAB and ROS simulation.The experimental performances of different path planning methods on real robot in the laboratory environment are compared to stress the improvement introduced by the algorithm developed in this thesis.