THESIS
2021
1 online resource (xii, 52 pages) : illustrations (chiefly color)
Abstract
This thesis presents an autonomous robotic system for pick-and-place manipulation that can handle objects even with a thin profile in a cluttered workspace. The system delivers a complete picking pipeline, which incorporates state-of-the-art manipulation technique named dig-grasping for simultaneous picking and singulation of objects from a dense clutter, and also another picking approach called scooping to compensate for the limitation of dig-grasping. Cutting-edge dexterous manipulation skill of ungrasping is also implemented to place the object in a quasi-static manner. The hardware designed to facilitate these manipulation primitives will also be presented along with the manipulation process. Picking failures and misalignment between picked object and the gripper finger are detected...[
Read more ]
This thesis presents an autonomous robotic system for pick-and-place manipulation that can handle objects even with a thin profile in a cluttered workspace. The system delivers a complete picking pipeline, which incorporates state-of-the-art manipulation technique named dig-grasping for simultaneous picking and singulation of objects from a dense clutter, and also another picking approach called scooping to compensate for the limitation of dig-grasping. Cutting-edge dexterous manipulation skill of ungrasping is also implemented to place the object in a quasi-static manner. The hardware designed to facilitate these manipulation primitives will also be presented along with the manipulation process. Picking failures and misalignment between picked object and the gripper finger are detected and addressed autonomously by the system to ensure the successful placement of object. The system is also capable of evaluating the placing accuracy by itself, and the data collected can be used to improve the precision of the placing process.
We demonstrate the application of the system in a Go playing scenario, which includes bin picking from a bowl of stones and subsequent placement on the Go board. The object for picking and the possible placing positions are detected autonomously in the process.
Post a Comment