Grasping force optimization approaches for common anthropomorphic grasps



Journal Title

Journal ISSN

Volume Title



A smart choice of contact forces between robotic grasping devices and objects is important for achieving a balanced grasp. Too little applied force may cause an object to slip or be dropped, and too much applied force may cause damage to delicate objects. The problem of determining appropriate contact forces is complicated by a variety of factors, including the nonlinearity of constraints and objective functions, the relatively large number of variables and constraints, and the need to compute an optimal solution in real time. Further, some methods are acceptable for real time computation for a small number of contact points but become unacceptable when the problem becomes more complex. This project examines some proposed approaches to grasping force optimization and compares their performance. An anthropomorphic model of the robotic hand is developed and is used to test grasping force optimization approaches using a variety of numerical examples representing tasks commonly performed by the human hand. Contact points between the hand and the object are predetermined, and several traditional approaches (nonlinear, linear, linear matrix inequalities, and neural networks) are tested for their computational efficiency, accuracy compared to the nonlinear method, and performance in response to a variety of external loads. Soft finger and point contact with friction contact models are both used; whole hand grasps and fingertip grasps are both examined. The robustness of each of these methods is then examined with respect to variability in both the contact points and the coefficients of friction using a probabilistic sensitivity analysis. Finally, a neural network is developed and proposed to improve the online performance of the optimization procedure.



Grasping force optimization