机构地区: 浙江师范大学
出 处: 《光学精密工程》 2014年第3期670-678,共9页
摘 要: 为了对轮式移动机器人(WMR)进行光滑、鲁棒、稳定的轨迹跟踪控制,分析了生物激励神经动力学原理,研究了非线性模型预测控制策略,提出了一种基于神经动力学思想的模型预测终端控制方法。首先,针对传统控制方法存在的初始速度跳变问题,利用神经动力学在信息处理方面的优良特性,设计了神经动力学控制模块;然后,根据模型预测控制原理给出了一个优化控制模块;最后,设计了终端域和线性反馈终端控制器来保证系统的全局渐近稳定性。仿真结果表明:利用所设计的控制方法进行曲线跟踪时,被控WMR系统收敛到参考轨迹的时间可从12s降到5s,初始线速度/角速度分别从[-3,4]m/s和[-5,6]rad/s缩小到[0,2]m/s和[-3,3]rad/s,且系统输出有界光滑,使WMR在完成轨迹跟踪的同时实现了全局渐进稳定。由于文中核心算法的推导过程不受WMR运动学模型限制,故该研究结论亦可应用于其他结构的移动机器人。 To track and control the trajectory of a Wheeled Mobile Robot (WMR) in the smooth, ro- bust and stable modes, the principle of the bio-inspired dynamics was analyzed, the nonlinear Model Predictive Control (MPC) was explored and a MPC approach based on bio-inspired dynamics was pro- posed. Firstly, a bio-inspired dynamics sub-controller was proposed based on the neuronsr excellent a- bility in information processing to overcome the velocity jump issue in the traditional control method. Then, an optimal sub-controller consisting of a cost function and four constraints was obtained based on the MPC principle. Finally, a terminal region and a terminal sub-controller were designed to stabi- lize the whole control system. Simulation results with the proposed control method indicate that the converge time of the WMR system to the reference trajectory has reduced from 12 to 5 s and the ran- ges of the initial linear and angular velocities are narrowed from [ 3, 4] m/s and [-5, 6] rad/s to [-0, 2] m/s and [-3, 3] rad/s, respectively. The output of the system is smooth and bounded, fulfil ling global asymptotic stability as well as higher tracking precision. As the algorithm used in deriva tion is not be limited by the WMR kinematics model, it can be used in other types of mobile robots.
关 键 词: 轮式移动机器人 模型预测控制 神经动力学 跟踪 稳定性
领 域: [自动化与计算机技术] [自动化与计算机技术]