Fuzzy Sliding Mode Joint Impedance Control for a tendon-driven robot hand performing peg-in-hole assembly

Autonomous robot hand manipulation has attracted a lot of attention recently. Especially when the robot fingers reach the object, each finger requires the coordination between the position and the force control in free and constrained space. It is not guaranteed to be stable when switching between different control modes. Furthermore, after grasping the object, we also want the system to have the ability to reduce the impact force when performing assembly tasks. For this purpose, instead of controlling the position and the force separately, we proposed a technique to control the characteristics of the interplay between the two, which is denoted by compliance. In this paper, we developed a control technique for peg-in-hole assembly combining visual and force feedback to enhance the compliance control of the object. The object space compliance can be transformed into the finger joint space, and the proposed Fuzzy Sliding Mode Joint Impedance Controller (FSMJIC), can achieve the desired finger joint compliance, deal with the model uncertainties as well as reject the external disturbances. It also solved the chattering problem of the conventional sliding mode control design. Once the joint impedance is achieved, the object compliance can be guaranteed. Both the real and the virtual force feedback were utilized in FSMJIC to enhance the performance. The virtual force was generated by using the concept of Visual Impedance, which can further reduce the impact force between the object and the obstacle. The 5-dof tendon-driven robot fingers were used as our testbed. The experimental results demonstrated the feasibility of our approach.