Redundant arm positioning control of the humanoid robot by linear visual servoing

This paper proposes a method to control the 4 d.o.f. arm of the humanoid robot by linear visual servoing. Linear visual servoing is based on the linear approximation between binocular visual space and joint space of the arm of the humanoid robot which has a similar kinematic structure as a human being. It is very robust to calibration error, especially to camera angle errors and joint angle errors, because it uses neither camera angles nor joint angles to calculate feedback command. Although we showed that 3D linear visual servoing is effective in positioning control of the 3 d.o.f. arm of the humanoid robot, the possibility to control the arm with more degree of freedom is not shown yet. In this paper, we propose a method to control 4 d.o.f. arm of the humanoid robot by linear visual servoing. We obtain the linear approximation matrix of 4 d.o.f. arm and express it as a function of the elbow joint angle by using the neural network. Some experimental results are presented to demonstrate the effectiveness of the proposed method.