The ultimate goal of robotic grasping and manipulation research is to obtain the same grasp stability and manipulation dexterity as human beings' hands. This thesis intends to solve some basic problems in robotic grasping and manipulation to make a step forward towards that goal....[ Read more ]

The ultimate goal of robotic grasping and manipulation research is to obtain the same grasp stability and manipulation dexterity as human beings' hands. This thesis intends to solve some basic problems in robotic grasping and manipulation to make a step forward towards that goal.

In this thesis, we develop a thorough study on the grasping force analysis and optimization of a whole hand grasp. We decompose the contact force space into four orthogonal subspaces, each with a clear physical interpretation. Based on linear matrix inequalities (LMI's) representations of the grasping constraints, we address and formulate the active force closure and the active grasp feasibility problems as LMI feasibility problems. We further propose a new cost index for the whole hand grasping force optimization problem, and formulate and solve it as a convex optimization problem involving LMI constraints.

We also proposes a novel modeling framework for the multifingered manipulation with finger gaits. Based on the different roles that the fingers play in a finger gait, they are classified into grasping fingers and free fingers. With this classification, an alternate representation of a grasp is introduced. With the consideration of both its discrete and continuous characteristics, the kinematics model of a k-fingered manipulation with finger gaits is formulated into a hybrid automaton. Several examples and simulation results verify the validity of the proposed modeling framework.

Based on the proposed modeling framework, we further develops a randomized manipulation planning algorithm to solve the finger gait planning problem. Considering the discrete and continuous topology of the automaton, both the discrete metric and continuous metric are defined on the hybrid state space. Based on the improved rapidly-exploring random tree (RRT) structure, a multifingered manipulation planner is proposed to find a feasible finger substitution gait. Simulation results on numerical examples verify the validity and performance of the proposed finger gait planning algorithm.