The design of an output-based robust disturbance rejection controller, aimed to solve the state tracking for the articulations of an experimental biped robot, was the main outcome of this study. The robust disturbance rejection controller included an auxiliary hybrid observer entailed to recover the angular velocity for each articulation. The estimated states served to perform the approximation of disturbances and non-modeled parts in the biped robot dynamics by implementing an extended state observer structure. The observer used the tracking position errors as input information, as well as considering the limb articular constraints, which are natural for biologically inspired biped robots. The effect of state constraints motivated the implementation of a hybrid observer with saturated output error injection. The controller design used the estimation of constraint velocity for solving the design of a tracking trajectory control to resolve the reproduction of the gait cycle by the bipedal robotic system. The Lyapunov stability theory served to obtain the laws which adjust the observer gains as well as to prove the ultimate boundedness of the tracking error as well. The evaluation of the suggested controller was realized on a numerical representation of the biped robot. These simulations illustrated the tracking performance of the hybrid robust disturbance rejection controller for all biped robot articulations in a decentralized structure. Experimental evaluations were also considered to validate the robust disturbance rejection controller design. A fully actuated biped robot was constructed and controlled by the robust disturbance rejection controller. The tracking results obtained by the robust disturbance rejection controller (in both the numerical and experimental evaluations) overcame the classical approach performances of diverse controllers as state feedback (proportional-derivative form) and regular robust disturbance rejection controller which did not consider the articulation constraints.

本研究的主要成果是设计一种基于输出的鲁棒干扰抑制控制器，该控制器旨在解决实验性Biped机器人的关节状态跟踪问题。鲁棒的干扰抑制控制器包括一个辅助混合观测器，该观测器必须恢复每个关节的角速度。估计状态用于通过实现扩展的状态观察器结构来对Biped机器人动力学中的干扰和非模型化零件进行逼近。观察者将跟踪位置误差用作输入信息，并考虑了肢体关节的约束，这对于受到生物学启发的两足动物机器人来说是很自然的。状态约束的影响激励了具有饱和输出误差注入的混合观察器的实现。控制器设计使用约束速度的估计来解决跟踪轨迹控制的设计，以解决双足机器人系统对步态周期的再现。李雅普诺夫稳定性理论的作用是获得调整观测器增益的定律，并证明跟踪误差的最终有界性。对建议控制器的评估是在两足动物机器人的数字表示上实现的。这些仿真说明了分散结构中所有Biped机器人关节的混合鲁棒干扰抑制控制器的跟踪性能。还考虑进行实验评估以验证鲁棒性干扰抑制控制器的设计。鲁棒的干扰抑制控制器构造并控制了一个全驱动的Biped机器人。