Cross-coupling control of dexterous hand based on neuron networks

In this paper, the author proposes a cross-coupling control approach based on neuron network, of which multiple parameters can be adjusted on line. According to the demand on the convergence of synchronization error in cross-coupling control, synchronization error item is introduced in this method to improve the performance function of BP (back propagation) neuron network, therefore it decreases synchronization error and meanwhile realizes the real-time adjustment of multiple parameters. The results of experiments show that compared with cross-coupling control method of which the parameters are stable, the method proposed in the paper holds the advantage that it can decrease position error and increase the stability of position tracking.