中文摘要：

提出了一种基于SR-UKF的主动状态建模方法用于移动机器人的在线故障检测和容错跟踪控制.通过对履带式机器人常见滑动故障的运动学分析,建立了带未知滑动故障参数的机器人运动学模型,并采用SR-UKF非线性滤波方法来联合估计机器人的位姿和滑动参数,在对机器人进行实时定位的同时实现了对快速变化（或突变）的滑动故障的在线跟踪和检测.在此基础上,将估计得到的自适应参数模型与基于Lyapunov分析的反馈控制律设计方法结合,获得了一致渐近稳定的轨迹跟踪控制结果,实现了针对在线故障自适应模型的容错控制重构.针对典型的阶跃式滑动故障参数变化的轨迹跟踪仿真实验表明了该方法的有效性.

英文摘要：

A square-rooted unscented Kalman filter（SR-UKF） based active state model is proposed for online fault detection and tolerant tracking control of mobile robots.The common slipping fault of a tracked vehicle is considered and the corresponding kinematic model with unknown slipping parameters is created.The slipping parameters as well as the pose of the robot are estimated simultaneously by the joint estimation framework to realize the real-time localization of robots and online identification and detection of the fast changing（or even abrupt changing） slipping fault.The obtained adaptive fault model is then incorporated to the tolerant control configuration in which a Lyapunov based feedback law is designed to achieve uniformly asymptotic stable trajectory tracking control.Simulation experiments for trajectory tracking with typical step changing slipping faults show the efficiency of the proposed method.