Khalid Mirza

Control of force distribution for power grasp in the DIGITS system

A linear programming approach is used to formulate and solve the force distribution problem in power grasps. The basic model used is taken from the DIGITS Grasping System, which consists of four three-degree-of-freedom fingers. The program solves for the required joint torques with limits imposed on their maximum value along with limits on contact normal reactions while optimizing an objective function. Friction constraints at the contacts are also included in the formulation to study their effects. The weight vector rotation method for analyzing power grasp stability is presented, which gives the maximum force the grasp can withstand in any given orientation. Power grasps with multiple contacts are found to be much more stable than fingertip grasps, with identical limits on the available applied torques. The load bearing and the torsional resistance capability of a wrap-type power grasp is studied for various size objects. Results show a significant increase in the maximum weight handling capability for completely enveloping type power grasps, and this improves even further with increasing friction at the contacts.<<ETX>>

General formulation for force distribution in power grasp

A general formulation of the force distribution equations for three-dimensional power grasp is presented. It allows for any number of contacts on the finger surfaces and the palm. The formulation not only includes the active forces and moments applied at the contacts, but also the passive forces resulting from frictional and geometric constraints, such as wedging effects. A grasp matrix is developed for the power grasp case and a new approach is taken to define the hand Jacobian matrix. Contact conditions are modeled to consider both constrained and unconstrained directions for the forces. A stability analysis for power grasp is performed in order to evaluate its stability properties. Due to the inherently stable nature of power grasps, the main objective is not the determination of optimum contact locations, but is the determination of joint torques for optimum force distribution. The stability region is defined and used as a measure to study and compare the different stability aspects of power grasps. Results of stability analysis for power grasps, using the DIGITS Grasping System as a model, are also provided.<<ETX>>

Dynamic simulation of enveloping power grasps

A dynamic simulation method for enveloping power grasps is presented. The method very effectively models all major grasp conditions and phenomena to achieve a realistic simulation of power grasp. A compliant contact model is developed to solve the static indeterminacy problem in a multiple-contact power grasp. A dynamic simulation algorithm is developed based on a discrete-time, state variable approach. The algorithm models friction, rolling, slipping, and wedging effects, as well as the changing topological structure of power grasp dynamics. The simulation algorithm is implemented for the DIGITS Grasping System. Results of quasi-static and dynamic stability experiments are discussed.<<ETX>>

Neural network control of force distribution for power grasp

The implementation of an artificial-neural-network (ANN)-based power grasp controller is discussed. Multiple points of contact between the grasped object and finger surfaces characterize power grasps. However, modeling is especially difficult because of the nature of the contacts and the resulting closed kinematic structure. Linear programming was used to train an ANN to control the force distribution for objects using a model of the DIGITS grasping system. Force control is implemented to insure that the maximum normal force applied to the object at the contacts is set to a prespecified level whenever possible. The ANN was able to learn the appropriate nonlinear mapping between the object size and force levels to an acceptable level of accuracy and can be used as a constant-time power grasp controller.<<ETX>>

Power grasp force distribution control using artificial neural networks

In this article, methods for force distribution control of power grasp are developed. A power grasp is characterized by multiple points of contact between the object grasped and the surfaces of the fingers and palm. The grasp is highly stable because of form closure. However, modeling power grasps is difficult because of the resulting closed kinematic structure and the complexity of multiple contacts. The first method used to obtain the desired force distribution is based on linear programming. In particular, a model of the DIGITS grasping system, under development at The Ohio State University, is used, and constraint equations are formulated for force balance and actuator torque limits. Supervisory control of the desired forces at the contacts is achieved by prescribing a desired clinch level. The objective function is designed to achieve the desired clinch level, except in cases where the specified force is inadequate to stably hold the object. Although this method yields the desired force distribution, a second method based on artificial neural networks (ANNs) is developed to achieve constant-time solutions. Linear programming solutions are used to generate training data for a set of ANNs. Two techniques, modular networks and adaptive slopes, are also developed and employed in the training to improve the training time and accuracy of the ANNs. The results show that the ANNs learn the appropriate nonlinear mapping for the force distribution and provide stable grasp over a wide range of object sizes and clinch levels.

A neural network interface to the DIGITS Grasping System

A neural-network-based interface between an operator and the DIGITS (dexterous integrated grasping with intrinsic tactile sensing) grasping system is proposed, and the initial results of the network training are presented. The neural network is responsible for accepting the description of an object to be held in a power grasp, and mapping these data into a set of actuator torques which will allow DIGITS to firmly grasp the object. The network should attempt to maximize the normal forces on the object to provide the best possible grasp while not exceeding a set level provided by the operator. The backpropagation neural network was trained with various quantities of hidden nodes and learning rates and then tested for stability and error with respect to the optimal solution. Useful results concerning the effect of learning rate and number of hidden nodes were obtained, as well as results indicating that the network can accurately determine torques for both trained and untrained objects