Perception and motion planning for autonomous grasping

Autonomous grasping is an active research area primarily because of the wide range of applications including manufacturing, logistics, assistive systems, and healthcare. Grasping and manipulation of objects in real-world environments is a challenging research problem, when properties of the object such as weight, shape, dimensions, stiffness etc. are unknown. This research work aims to explore sensor fusion to help estimate the object properties and then plan a stable grasp based on the manipulation requirements.

Preliminary results on autonomous pick-and-place task through the use of artificial fiducial markers with a three finger Robotiq gripper is shown below:

Click here for video