Aiming at the multifingered grasping model under the condition of point contact with friction, the concept of contact safety margin is proposed according to the experience of human grasp. The nonlinear optimization model of internal force with contact safety margin is built by modifying the constraint condition of friction cone, and the corresponding contact force can therefore be obtained. In order to obtain the optimal grasping pose, the relationship between joint torque and contact force is established by taking object pose parameters as variables, and the multi-objective optimization model of grasp is built based on the performance indices including position-grade and relative loading capability in the joint space of robot hand. Finally, a three-fingered robot hand grasping object is calculated and analyzed, and multi-objective particle swarm optimization algorithm is applied to plan the object pose and a set of non-dominated solutions is also obtained. The results demonstrate that the proposed method can efficiently improve grasping performance under the condition of safe grasping.