Intelligent control of object acquisition for power grasp

An intelligent control architecture is developed for a robotic grasping system capable of acquiring an object into a fully enveloping power grasp. Control of the internal forces of the grasp is provided, along with trajectory control of object position as the object is picked up. Fuzzy control techniques are used for control of internal forces in the power grasp, and an artificial neural network (ANN) provides a means of in-process nonlinear friction estimation. The authors show that inclusion of the ANN improves the tracking accuracy of the object position. The architecture is described, and several experimental studies, using a simulation model of a simple two-link system, demonstrate performance gains realized by using in-process friction estimation