Two arm adaptive load apportioning using a Hopfield neural net

The interactive control issue of two robot manipulators jointly grasping a rigid object with no slippage is addressed. The use of wrist force/torque sensors on each robot arm is required to implement this control strategy. There are two items to control: (1) the forces/torques applied by the two arms that cause internal stress in the object (bias forces); and (2) the forces/torques applied by the two arms that cause acceleration of the object and overcome gravity (inertial forces). Since an object in 3-space has 6 degrees of freedom (motion along the three axes and rotation about the three axes), the bias forces or apportioned inertial forces in each of these 6 degrees of freedom can be selectively controlled. Unfortunately, it is not possible to do both in a single direction. The Hopfield neural network is used to determine the optimum (in a weighted least squares sense) load-apportioning feedback gains. Simulation results are presented