Postural Synergy Analysis Based Underactuated Prosthetic Hand Design And Myoelectric Control Method

The birth of myoelectric prosthetic hand has evolved prosthesis from a machine fixed to human body to a motor function restoration device. However, current commercial myoelectric prosthetic hand with single degree of freedom(DOF) is still far from the ultimate goal of reconstructing one’s motor function. Many countries have launched several research projects for developing advanced prosthesis.Versatile multi-DOF myoelectric prosthetic hands are often desired for such developments. Two major challenges shall be addressed: i) the prosthetic hand should be dexterous enough to replicate the delicate movements and operation functions of a human hand; ii) the prosthetic hand should be connected to the one’s motor system with the motion intentions correctly and smoothly interpreted. This dissertation focuses on advancing both aspects to enable the development of a versatile prosthetic hand controllable via a myoelectric interface.Study in neurology shows that CNS(Central Nerve System) controls hand muscles for different postures in a coordinated manner. The coordination is referred as to a postural synergy. Implementation of postural synergies is considered a promising way to achieve dexterous control of a multi-DOF prosthetic hand using limited bandwidth from a myoelectric interface. This dissertation proposes models of hand kinematics and grasping statics for an underactuated prosthetic hand with postural synergies and jointlevel compliance. Quasi-static grasping process of the underactuated hand is analyzed and solved. The proposed model and methods are compared with existing ones to demonstrate the improvements, and applicability. A grasp quality measure and grasp planning methods are also proposed and investigated.A simulation system named “SJT-Grasp Sim”is established based on aforementioned grasping studies. It can be used for design, simulation, analysis and optimization of the similar prosthetic hands. The simulation system is elaborated with numerical examples in detail. It can improve the design efficiency and shorten the development cycle.A postural synergy based underactuated hand prototype called ”SJT-5” is developed. It has 18 DOFs and anthropomorphic form. A design method of this type of underactuated hand is proposed. First different grasps are performed on a fully-actuated robotic hand, and postural synergies are extracted. The grasping performance is evaluated in the simulation system. Then underactuated grasping is verified using a fully-actuated hand with digitally implemented postural synergies. Mechanical implementation of the postural synergies is investigated. Extracted postural synergies are used to design the transmission unit directly. The experiments show that the robotic hand can achieve various daily-life grasps effectively.Examining the performance of an EMG-based control interface using pattern classification, the dissertation proposes a proportional EMG signal decoding method. The performance of this proportional EMG control interface was evaluated quantitatively through a series of experiments, and compared to the performances of other humanmachine interfaces to demonstrate the improvements. Finally, it is utilized to control the SJT-5 Hand to perform various grasps. The results show that this proportional EMG interface is able to command the postural synergy based underactuated prosthetic hand well for versatile grasps.