Motion planning using a ToF camera for manipulation in cluttered environments

Time-of-Flight camera can help a service robot to sense its 3D environment. The modeled environment information can be used by motion planning algorithms to plan a collision free movement for the robot, or by grasp planning algorithms to find feasible grasps for an a priori unknown object in the scene. To guarantee that the models used by the planning algorithms are consistent with the real world, calibration of the Time-of-Flight camera is necessary. In this article the intrinsic and extrinsic parameter calibration procedures applied in the DESIRE project are presented. A segmentation using GPU calculation is used to seperate the sensed model and a priori information of the environment. We use a motion planning algorithm based on Probabilistic Road to plan collision free movement for DESIRE platform.