Novel design and motion planning strategies for palletization tasks using collaborative robots in automated warehouses

This dissertation introduces novel design and motion planning strategies for collaborative robots in automated warehouses. Traditional warehouses are characterized by human-centred industrial processes where tasks are performed manually by human operators. Automated warehouses were invented to improve efficiency when advanced autonomous robot systems became available. The objective of this thesis is to investigate novel approaches for manipulation tasks in automated warehouses, such as palletization and depalletization of single packages using collaborative robots. Collaborative robots have been the focus of attention due to their flexibility and their ability to work in shared workspace with human operators. The contributions of this dissertation range from developing a prototype platform to implementing novel algorithms. Hardware devices available on the market, such as collaborative robots and vacuum grippers, are compared. In particular, collaborative robots are compared in simulated depalletization tasks, while vacuum grippers are evaluated through manual experiments conducted by a human operator. Based on the comparison results, a prototype platform for depalletization tasks was developed, featuring a collaborative manipulator mounted on a mobile platform. In addition to the prototype platform, this thesis presents two novel strategies for object manipulation: a task and motion planning algorithm that uses a probabilistic model to estimate the feasibility of pick and place actions and a procedure for estimating the displacement error of grasped boxes by inducing collisions between the grasped box and a fixed fixture in the environment.