Force Planning And Synchronized Control Of Muilti-finger Hand

Robotic system equipped with multi-finger hand has great potential in the field ofcomplicated manipulation tasks. Research on control of multi-finger hand is of greatsignificance to enhance the performance of the entire robotic system. According to thenational nature science fund program “Research of the real-time force optimization andindependent manipulating control of the multi-finger robot hand”(No.60675045), thisdissertation studies the control scheme of cross-coupling, impedance force tracking andgrasping force optimization of multi-finger hand with purpose to improve manipulationaccuracy.Precise manipulation of multi-finger hand is done by fingertips. As a finger isessentially a series robot, motion of fingertip is actually coupled by all finger joints.Therefore, accuracy of fingertip motion is related to each joint’s accuracy whileconventional control scheme ignored this relationship. This dissertation discussesaffection to fingertip motion accuracy by different forms of cross-coupling error,synchronizing error among finger joints in the form of percentage is selected to becoupled into finger control loop. System stability will be affected with cross-couplingerror introduced, thus a trajectory tracking controller with coupling compensation isdesigned, and the controller is proved to be stable. Experiments are demonstrated to testthe controller and to analyze the affection of coupling parameter on fingertip motionaccuracy and joints’ synchronizing errors.Multi-finger hand applies force to operation object during process of manipulation,and impact may occur when finger transits from free space to constrained space. Tosolve this problem, this dissertation studies control scheme of impedance force trackingin Cartesian space. Cartesian impedance is converted to equal joint impedance so thatCartesian impedance control can be achieved in joint space. To achieve precise forcecontrol, an indirect adaptive impedance force tracking controller with an inner positioncontrol loop of cross-coupling controller is designed after a reference force is introduced.Experiments are demonstrated to test the controller and analyze the effect of differentinner position control loop.Precise manipulation requires grasping force of multi-finger hand to be largereasonably, as well as to balance external force and to satisfy friction cone constraints.Therefore, it is essential to optimize grasping force when force is planned. Moreover,practical application requires force optimization to be instant. This dissertation studiesforce optimization algorithm of multi-finger hand. Based on gradient flow algorithm,linear combination algorithm is designed to compute initial grasping force, thealgorithm is proved to be proper theoretically. A grasping example is constructed tosimulate the algorithm designed and analyze the effect of weight value in cost function. Integrated experiments results indicate controllers and algorithms designed in thisdissertation are effective.