Research on Disturbance of Upright Balance of Biped Humanoid Robot Based on AWPSO-LQR

摘要

On the premise of ensuring the safety of the biped humanoid robot, this paper focuses on the issue of dynamic balance anti-interference in its upright state. Considering the impact of model accuracy on system performance, it is abstracted as a seven-link model, and the multi-body dynamic model of the single-leg support phase in the sagittal view is established from the energy point of view. And then, based on the above model, a control architecture-full state LQR controller is designed. On the basis of the traditional particle swarm algorithm(PSO), the adaptive inertia weighting scheme based on the success rate of updating the particle velocity, that is, the adaptive weighted PSO(AWPSO) is used to determine the optimal weight matrix of LQR. The simulation experiment results show that the controller can restore the biped humanoid robot from a variety of different initial postures and stabilize it near the upright balance position in a short time, it has a wide dynamic range and does not need to change the controller parameters, the algorithm complexity is also considerably reduced. In addition, our controller has good suppression ability for instantaneous interference.