Content area
Abstract
This thesis presents a contribution to the kinematic analysis, optimization and trajectory planning of articulated hands. These multifingered grippers are composed of open-loop and closed-loop kinematic chains which leads to a good stiffness and accuracy. The study reported here addresses the kinematics of planar and spatial articulated hands including tree-structure mechanisms and complex mechanical systems. To this end, geometric and kinematic concepts developed for manipulators are applied to articulated hands.
The direct kinematic problem for this class of mechanisms is obtained from the kinematic sub-chains and leads to nonlinear algebraic equations. Furthermore, the inverse kinematic problem leads to an infinity of solutions for redundant hands. However, in order to restrict the number of solutions of the inverse problem, a kinematic model is derived in this thesis which allows one to obtain the joint variables by specifying the position and orientation of the end-effector. The formulation leads to a polynomial solution for the planar case and to nonlinear algebraic equations for the spatial case.
The workspace of multifingered grippers is formulated from the kinematic equations of the direct problem. Geometric descriptions of the workspace have been obtained from the intersection of circles in the plane and spheres in 3D space. The area or volume of the workspace is computed using the Gauss Divergence Theorem which leads to an efficient implementation. The optimization of the actual workspace is then undertaken using some performance criteria based on the intersection between the workspace and a prescribed workspace. Optimum solutions are obtained for planar and spatial cases.
The optimum kinematic design of multifingered grippers is formulated as an optimization problem. To this end, the dexterity concept is chosen as a performance criteria. The analysis is based on the determination of all the configurations and architectures leading to a minimum condition number. Isotropic or quasi-isotropic configurations are obtained for particular classes of articulated hands.
The optimum design of multifingered grippers is also formulated as a global dexterity maximization problem. Indeed, the properties of the jacobian matrix associated with the articulated hands are exploited to obtain analytical or numerical expressions of the global dexterity. Optimum solutions are obtained analytically for particular classes of articulated hands.
Using the algorithm from Liegeois, the kinematics of articulated hands in the presence of redundancies is formulated as a local dexterity maximization problem, and smooth joint trajectories are obtained. Finally, the optimum contact conditions between the object and the fingers during the motion are used to optimize the architecture of the grippers for maximum grasping stability. Optimum results are obtained in the planar and spatial cases.