Online Motion Planning for Humanoid Robot Based on Embedded Vision System

In this paper, we present a control system based on embedded system according to the features of humanoid robot. First, the image captured by the vision system is recognized through the improved look-up table method. And then with a smooth gait planning of turning motion for humanoid robot presented, the next several motions for the robot can be proposed by the local motion planner based on the minimum of energy dissipation. After that, a control decision is made for the humanoid robot to plan the motions by the method of gait generation off-line and adjustment on-line according to the rules of marathon held in FIRA. In this way, simulations for image processing and experiments on real humanoid robot HIT-2 are present.