Collision-free path planning for industrial robot applications

Nowadays, robotics is growing rapidly thanks to its huge flexibility and potential in many industrial applications. One of the recurring problems is the collision-free path planning that allows the robot to move from an initial pose to a final pose safely, avoiding all objects in the scene. In COMAU, common planners belonging to the sampling-based category are used for collision-free path planning. To do this, the MoveIt framework is used in which is present the OMPL library that contains many of the sampling-based planners. These planners are fast and effective even in complex environments, however they generate geometric paths that are not optimized and do not consider kinematics constraints. Indeed a second step is needed, performed by the C5G controller on board robot, which also considers kinematics constraints and adds a timing law to the collision-free geometric path in order to obtain the final robot trajectory. In this thesis work, many types of path planners present in the state of the art are analyzed. Then it is decided to develop a comparison between the performances of OMPL in MoveIt currently used in COMAU and TrajOpt which is an innovative optimized-based planner already present in the Tesseract framework. TrajOpt plans collision-free trajectories with an optimized path. In this way the final paths are much better and smooth, and already respect the kinematics constraints. On the other hand, the higher computational complexity increases the planning times required. Thanks to this thesis work, it is possible to enhance the comparisons about the performances of TrajOpt and the OMPL planners in typical industrial robotics scenarios.