拟人机器人关节空间自适应H_∞位置跟踪控制

Adaptive H_∞ position control for tracking in joint-space of humanoid

拟人机器人的建模和控制一直是一种开放的、富于挑战性的问题,研究了一种正交轮式移动车为载体的拟人机器人控制问题。首先建立了系统模型,据此提出一种新的基于NN的自适应H∞位置跟踪控制器,将鲁棒非线性H$控制方法与模型的直接自适应神经网络技术作了自然集成。然后证明了其鲁棒稳定性,并进一步分析了控制器中的重要细节。最后仿真研究验证了该控制器的正确性和有效性。

How to model and control a humanoid is always an open and challenging subject.This paper describes modelingand controller design for a humanoid with a holonomic omnidirectional platform consisted of three lateral orthogonal wheelassemblies.The system model is constructed.And a new neural-network-based adaptive H∞ position tracking controller is here-by derived.The robust nonlinear H∞ control approach and model＇s direct adaptive neural network technique are integrated to-gether naturally.Robust stability is proved.Furthermore,the key details of the controller are analyzed.A simulation study forthe humanoid＇s vehicle and torso is given.Simulation results indicate that the proposed controller is valid and effective.