Research On Grasp Planning And Optimization Of Grasping Forces For Robotic Dexterous Hands

With the constant development of economy, the increasing constantly of product category, the traditional grippers installed at the end of manipulator can not gradually meet the manufacturing requirements, the robotic dexterous hands as the universal end operator, will play an important part in the moden industrial production field.In the extreme or dangerous working environment such as outer space, underwate and nuclear radial field, the robotic dexterous can carry out the dangerous and complex tasks instead of human being, which has great potential application prospects.In the field of entertainment and service, the dexterous hands assembled in the humanoid robot are paid more and more attation. Grasp planing, decomposition of constant force and planing of defective grasp, the optimization of grasping forces and the hydrodynamic modeling of the finger are reseached deeply in the paper.Based on the grasping static and kinematic equation, the general solution of joint torque is gotted by using the weighted generalized inverse, indicators of the volume of electrical external force ellipsoid, the flattening of electrical external force ellipsoid, the volume of electrical contact internal force ellipsoid and the flattening of electrical contact internal force ellipsoid, which reflect the transfer from the joint torque to the external force and contact internal force in all directions, are brought forward,then the optimal grasping indicator--the maximum volume of electrical external force ellipsoid and the minimum volume of electrical contact internal force ellipsoid in the accepted flattening and the grasp planning method based on task are put forward, the efficient of the grasp planning indicator and the feasibility of the grasp planning method are tested by a grasping example at last.Based on the decomposition of constant force subspace, the constant force is decomposed into the joint torque and the passive balanceed external force definitely, the new evaluation indicator of defective grasp stability--the maximum passive balanceed external force is present, and the general formula of the maximum passive balanceed external force is derived by decomposition of constant force above, the defective grasp indicator of maximizing the passive balancced external force is raised,and the optimized planning on grasping a disk with two fingers is completed with the indicator above as the objective function.With the linear single polyhedral cone, paired polyhedral cones and ellipsoidal polyhedron taking the place of the non-linear single friction cone, paired friction cones and friction ellipsoid respectively, the linarization of point constant with friction and two kinds of soft-finger constant is completed, and the relationship between the linarization error of frictionrestraint and the faces of the polyhedron is derived, the non-linear grasping constraints are simplified to linear constraints with joint torque as variable, the optimization model with linear constraints as optimization constraints and with the minimum sum-of-squares of joint torques as objective function is set up,and the timely and effective algorithm of the active costraintset method is carried out at last.Based on the hydrodynamics the effect factors of water environment to the dynamic model of the dexterous hands underwater are gotted, the computing method of the hydrodynamic torque, the added mass torque,water impact torque and the buoyancy torque suffered by the joints underwater is carried out, the the exact dynamic model of finger underwater is built up,then the hydrodynamic equation of the typical side-sway finger with three joints is deduced and the dynamic simulation is completed,the incidence of all kinds of factors to the finger is carried out at last.The experimental system of the dexterous hand is set up, and the experiments of the joint motion control, the position control of a single finger, optimization of grasping force and grasp planing method are completed, the exeriment results show that the multi-finger dexterous hand can complete the exact position control in the joint space and Cartesian space, objects of different shapes can be grasped in different grasping ways, with the grasp planning method based on the volume and flattening of electrical force ellipsoid, the optimal grasping configuration and the position of the constant points can be achieved.