Christian Scheurer

Path planning method for palletizing tasks using workspace cell decomposition

We present an algorithm to solve path planning tasks for industrial 6-axis robots with a focus on palletizing and handling applications. Given a 3D environment our algorithm decomposes the Cartesian workspace of the robot into 3-dimensional cylinder slices centered around the robot base. The resulting decomposition is searched for a sequence of collision-free cells connecting the start and goal position to get a smooth joint trajectory that can be executed by a robot controller and totally lies inside this cylinder cell sequence. We compare our algorithm to an existing, random-based planning algorithm using different scenarios.

Parallel cartesian planning in dynamic environments using constrained trajectory planning

We present a parallel Cartesian planning algorithm for redundant robot arms and manipulators. We pre-compute a roadmap, that takes into account static obstacles in the environment as well as singular configurations. At runtime, multiple paths in this roadmap are computed as initial trajectories for an optimization-based planner that tends to satisfy various constraints corresponding to demands on the trajectory, including end-effector constraints, collision-free, and non-singular. We highlight and compare the performance of our parallel planner using 7-DOF arms with other planning algorithms. To the best of our knowledge, this is the first approach that can compute smooth and collision-free trajectories in complex environments with dynamic obstacles.