Object Manipulation for a Humanoid Robot Based on an Integrated Control Strategy

When the humanoid robot manipulates an object, the position of the humanoid waist has to be adjusted in real-time to make the robot stable. The position change of the humanoid robot waist will make affect to object manipulation based on visual servoing. However, this issue has not been studied in previous interesting actions. So, in this paper a control strategy that combining visual feedforward and visual feedback control is proposed for the reach-to-grasp task. Visual feedforward control simplifies the reach-to-grasp task, which uses a look-up table algorithm to get the position of the waist rapidly and the joint angles of the arm. Visual feedback control increases the robustness by compensating the errors caused by the weak calibration model. The rapidity and high robustness of the system is verified by the experiment results.