This thesis presents two novel robotic techniques respectively for picking and placing, which are applicable to relatively thin objects that can be quite challenging for robots to handle.

First, we propose a robotic picking technique, high-speed scooping, for rapidly picking thin objects off from the hard surface by leveraging the compliant interaction between the direct-drive gripper and environment. The technique is based on the scooping manipulation to obtain a secure pinch grasp for relatively thin objects that are lying on a support surface. We present the procedure of scooping manipulation featuring the stiffness control framework for a two-fingered compliant gripper to pick up thin objects at high-speed. An implementation of the proposed approach is also presented using our custo...[ Read more ]

This thesis presents two novel robotic techniques respectively for picking and placing, which are applicable to relatively thin objects that can be quite challenging for robots to handle.

First, we propose a robotic picking technique, high-speed scooping, for rapidly picking thin objects off from the hard surface by leveraging the compliant interaction between the direct-drive gripper and environment. The technique is based on the scooping manipulation to obtain a secure pinch grasp for relatively thin objects that are lying on a support surface. We present the procedure of scooping manipulation featuring the stiffness control framework for a two-fingered compliant gripper to pick up thin objects at high-speed. An implementation of the proposed approach is also presented using our custom-built two-fingered direct-drive gripper.

Second, we present a robotic technique for secure and precision placing, named as dexterous ungrasping, which refers to the task of releasing a pinched object securely and accurately to the target surface through dexterous manipulation, that is similar to how Go player places a stone on the board with the index and middle fingers. A planning framework is presented to compute a secure ungrasping motion using the contact interactions between the gripper, object, and environment. Strategies of robustly executing the planned ungrasping motion are also devised by analyzing the possible failure cases and caging conditions.

Experiments in a range of scenarios demonstrate the effectiveness of both presented techniques. The high success rate of both picking and placing operations obtained from the experiment shows that our techniques are applicable to the problems in the industry of manufacturing and assembly.