Support-vector-machines learning controller based on small sample sizes for biped robots
一种小样本支持向量机控制器在两足机器人步态控制的研究

Conventional machine learning methods such as neural network(NN) use empirical risk minimization(ERM) based on infinite samples, which is disadvantageous to the gait learning control based on small sample sizes for biped robots walking in unstructured, uncertain and dynamic environments. To deal with the stable walking control problem in the dynamic environments for biped robots, we put forward a method of gait control based on support-vector-machines(SVM), which provides a solution for the learning control issue based on small sample sizes. A support vector machine regression(SVR) method for gait control with mixed kernel functions is proposed, and the proposed method shows superior performance when compared with SVR with radial basis function(RBF) kernels or polynomial kernels, respectively. Using ankle trajectory and hip trajectory as inputs, and the corresponding trunk trajectory which guarantees the ZMP criterion as outputs, the SVM is trained based on small sample sizes to learn the dynamic kinematic relationships between the legs and the trunk of the biped robot. Then the trained SVM is incorporated into the control system of the robots. Robustness of the gait control is enhanced, which is advantageous to realizing stable biped walking in unstructured environments. Simulation results demonstrate the superiority of the proposed methods.