Development and testing of docking functions in industrial settings for an autonomous mobile robot based on ROS2

This dissertation is the result of a six-months internship at G.D S.p.A. for the preparation of the thesis project. The final goal is to develop algorithms on the ROS2 framework that could be used to control an Autonomous Mobile Robot during the operations of detection and approach of a docking station with high precision, needed to operate a recharge of the AMR itself or some operation on the host machines. The automation of these operations ensures a substantial increase in safety and productivity within a warehouse or host machine lines since it permits to the AMR to work without requiring an operator for longer time or even to substitute the operator itself. The presented method uses both lidars and an onboard camera. The trajectory from the starting position to the approximate area of the docking station is computed using data obtained from the three lidars around the AMR body. The final approach is implemented by detecting an ARUCO code positioned on the dock assembly through a camera. A sequence of intermediate positions is defined according to the pose estimations, and then reached with a mix of standard navigation and a proportional position control in the very last part of the movement trajectory. The precision of the docking position turned out to have less than one centimeter error around the desired target, the orientation error is a fraction of a degree. The docking times vary based on how far the AMR is from the docking station, but the last phase of the procedure is always completed in around seventeen seconds. The solution is implementable and will be evaluated on the real platform in the coming months.