Use of Support Vector Machines: Synergism to Intelligent Humanoid Robot Walking Down on a Slope

Abstract

In this paper intelligent humanoid robot walking down on a slope with support vector machines is presented. Humanoid robots can be used as proxies or assistants to humans in performing tasks in real world environments, including rough terrain, steep stairs, and obstacles. But the dynamics involved are highly nonlinear and unstable. So the humanoid robot can not get the stable and reliable biped walking easily. As a significant dynamic equilibrium criterion, zero moment point (ZMP) is usually employed and we are establishing empirical relationships based on the ZMP trajectory as dynamic stability of motion. Support vector machines (SVM) are applied to model a ZMP trajectory of a practical humanoid robot. The SVMs’ performance can vary considerably depending on the type of kernels adopted by the networks. The experimental results show that the SVM based on the kernel substitution provides a promising alternative to model robot movements but also to control actual humanoid robots.