A connectionist model of human grasps and its application to robot grasping

Presents a connectionist model that can learn human grasping rules. The model domain is a taxonomy of manufacturing grasps. The input vector specifies task and object attributes such as dexterity and shape, and the model predicts the required grasp. Experiments show that the model is capable of learning grasping rules and predicting correct grasps for new objects that it has not been trained on. The authors finally discuss how this model can be used as part of a robotic system for grasping objects.

[1]  George A. Bekey,et al.  A strategy for grasp synthesis with multifingered robot hands , 1987, Proceedings. 1987 IEEE International Conference on Robotics and Automation.

[2]  W. Eric L. Grimson,et al.  Handey: A robot system that recognizes, plans, and manipulates , 1987, Proceedings. 1987 IEEE International Conference on Robotics and Automation.

[3]  Joe Jackson,et al.  Knowledge-based prehension: capturing human dexterity , 1988, Proceedings. 1988 IEEE International Conference on Robotics and Automation.

[4]  T. Iberall,et al.  Neural network architecture for robot hand control , 1989, IEEE Control Systems Magazine.

[5]  Mark R. Cutkosky,et al.  On grasp choice, grasp models, and the design of hands for manufacturing tasks , 1989, IEEE Trans. Robotics Autom..

[6]  Mark R. Cutkosky,et al.  Modeling manufacturing grips and correlations with the design of robotic hands , 1986, Proceedings. 1986 IEEE International Conference on Robotics and Automation.

[7]  Geoffrey E. Hinton.  Connectionist Learning Procedures , 1989, Artif. Intell..

[8]  Robert A. Jacobs,et al.  Increased rates of convergence through learning rate adaptation , 1987, Neural Networks.

[9]  Geoffrey E. Hinton,et al.  Learning distributed representations of concepts. , 1989 .