This paper describes a specific person recognition method and tracking control system with Kinect on-board a mobile robot. An individual recognition algorithm using individual skeletal features is proposed and the experiment results are reported in this paper. The experiment for verifying the proposed method was conducted by recognizing a specific person from two moving persons. The experiment results show that the proposed algorithm is effective in the case that there are individual differences between skeletons.

We analyse the mechanisms of human walking patterns and propose two design methods of neural oscillatory networks for generating various desired biped walking patterns of an 8 joints model in 3D plane. First, we discuss how the construction and the parameters of the network determine the inner state variable of each neuron in the network. Then we try to use two different methods to build the suitable network which can generate the desired walking patterns. The first method is to lower the system dimension to 3/spl times/n by considering some mechanisms in human walking and solve the connection weights of the network by the harmonic balance method. The second method is to calculate all kinds of connections of the network by genetic algorithms. From the results of simulations the methods were proved to be effective.