| A hand is considered as an agent of human brain and is the most intriguing and versatile appendage to the human body. Replicating the human hand in all its various functions is still a challenging task due to the extreme complexity of a human hand, which is 27 bones, controlled by about 38 muscles to provide the hand with 22 degrees of freedom (DOFs), and incorporates about 17,000 tactile units of 4 different types. This dissertation investigates control strategies using hybrid of hard control and soft computing techniques. First, the problems of forward kinematics, inverse kinematics, differential kinematics, and trajectory planning for a serial n-link joint prosthetic hand are addressed. Then, the dynamic equations of hand motion are derived via Lagrangian method. Based on the kinematic and dynamic models of the prosthetic hand, hard control (HC) techniques, such as feedback linearization, PD/PI/PID control, optimal control, and adaptive control, are proposed. A knowledge-based system is preferred to address the tasks where the nature of the problems and solutions are not well defined or not known. Therefore, this dissertation presents soft computing (SC) or computational intelligence (CI) techniques, including fuzzy logic (FL), neural networks (NN), adaptive neuro-fuzzy inference system (ANFIS), tabu search (TS), genetic algorithm (GA), particle swarm optimization (PSO), developed adaptive particle swarm optimization (APSO), and condensed hybrid optimization (CHO). Next, by fusion of HC and SC techniques, hybrid control strategies are developed to cash on the advantages of HC and SC. Finally, numerical simulations with a two-link thumb, and three-link fingers using fusion of PD controller and PSO, PID controller and GA, and optimal and ANFIS and GA, clearly demonstrate that the integration of SC and HC methodologies is superior to using either HC or SC technique alone. |
