Grasping control for a neuro-cognitive robot

Behavior learning via continuous time recurrent neural network (CTRNN) could facilitate the robotic manipulation planning and control. However, robotic behavior learning could not guarantee a precise manipulation for grasping. In our context, precise grasping refers to firm but delicate grasping motion. To generate firm and gentle grasping motion for a neuro-cognitive robot, the robot should have the ability to detect the forces exerted at its end effectors and react accordingly. The main objective of this paper is to propose a systematic framework to work out a grasping controller which enables a neuro-cognitive robot to implement precise pick, hold and place motions. The feasibility of the proposed controller has been verified via a soft cubic box grasping test by implementing on a neuro-cognitive robot, NECO-III.