Whole body motion control strategy of humanoid robot based on double-layer quadratic optimization

Humanoid robots have better adaptability and universal operation capabilities compared to other types of robots due to their structural and motion similarity to humans. However, their underactuated and multi-degree of freedom characteristics make control strategies more complex. This article proposes a double-layer quadratic programming strategy for a whole-body motion framework that decouples the two main factors affecting humanoid robot motion: ground reaction forces and joint acceleration. The first layer calculates optimal ground reaction forces that satisfy the ZMP constraint, centroidal momentum change, and friction constraint, while the second layer calculates full-body joint angle acceleration to complete various tasks, including base posture tasks and bipedal posture tasks. The novelty of this double-layer quadratic optimization method lies in the decoupling of stability and motion ability in humanoid robot motion and expressing them separately. The weight allocation is clear, and the relevant constraint weight can be adjusted to change the bias of the humanoid robot between the two, thereby adapting to the needs of different movements. Moreover, the quadratic optimization equation has a fast solution speed and can meet the real-time requirements of humanoid robots. Finally, we carried out two legs squat and one leg squat respectively in the simulation to verify the feasibility of this method, and the experimental results show that it can meet most of the task requirements.