Robotic Part Handling for Unstructured Industrial Applications, 10-R8301
Principal Investigator
Clay Flannigan
Inclusive Dates: 04/01/12 – 11/01/12
Background — Recent developments in 3D sensor technologies, adaptive robotic grippers and advanced perception and planning algorithms are rapidly advancing the use of robots in complex and dynamic environments. Traditional industrial application of robots requires that parts be precisely located using dedicated fixtures so that there is little uncertainty in the location of the workpiece and condition of the workspace. More advanced industrial robotic systems use 2D or 3D vision sensing to handle minor variations in part locations. However, there are large classes of problems where high part variability or dynamic environments prohibit rigid fixturing methods and confound commercially available vision solutions. An example application is sorting residential recyclables, where there is an almost infinite variability of parts to be manipulated. Aerospace manufacturing is another market that is challenged by high-mix, low-volume processes.
Approach — SwRI has previously developed an open-source software framework for industrial robotics called ROS-Industrial that leverages the huge community of robotic researchers using the Robot Operating System (ROS). This research is extending the ROS-Industrial program through investigation of perception, motion planning, and grasp planning methods to address unstructured manipulation. Specifically, we are using the Microsoft® Kinect sensor for colorized 3D range image acquisition. Object recognition and pose estimation algorithms are being developed using the Kinect data to identify objects in cluttered environments. Both vacuum and adaptive finger grippers are being employed for grasping. Motion planning objectives include the generation of efficient and collision-free motion even in dynamic or complex scenes.
Accomplishments — In the first six months of this 12-month project, the broad research objectives were narrowed by identifying end-user requirements and by creating a hierarchy for the problem domain. A robot cell has been integrated that includes a Motoman SIA-20 robot, Robotiq 2 and 3-finger grippers, vacuum grippers, and Kinect sensing. The robot cell has been integrated with the ROS-Industrial software and is capable of tabletop manipulation of randomly oriented objects using the vacuum gripper. A novel method has been developed for object recognition and pose-estimation based on object templates. This perception algorithm is robust for tabletop recognition using just shape cues (visual texture is not required). Future work will include continued research on perception methods for highly cluttered scenes such as those found in bin picking. Additionally, a demonstration capability will be created that permits quantitative benchmarking of the perception and manipulation approaches.
