自适应轨迹任务模仿的模仿学习方法研究  

Research on Imitation Learning Method of Self-adaptive Trajectory Task Imitation

作　　者：于建均 姚红柯[1,2] 左国玉 阮晓钢 YU Jian-jun;YAO Hong-ke;ZUO Guo-yu;RUAN Xiao-gang(Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China;Beijing Key Laboratory of Computational Intelligence and Intelligent System,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]北京工业大学信息学部,北京100124 [2]北京工业大学计算智能与智能系统北京市重点实验室,北京100124

出　　处：《控制工程》2021年第2期266-274,共9页Control Engineering of China

基　　金：国家自然科学基金项目(61773027);北京市教育委员会科技计划重点项目(KZ201610005010)。

摘　　要：机器人模仿的学习方法在行为运动的模仿上受到示教速度的限制,导致机器人模仿的速度也受到限制,无法更好发挥机器人的性能。为了提高机器人行为模仿的快速性,提出了一种自适应改变机器人模仿学习运动速度的方法。首先通过基于动态系统的方法建模示教运动,并将动态系统稳定的充分条件作为约束,确保行为模仿的稳定性。其次构造了一个随机器人状态到目标点的距离而变化的非线性函数,将非线性函数作为参数与系统模型结合,以便自适应地调整模仿的速度。最后给出了4种模仿学习评价的方法来评价模仿的性能。实验结果表明,提出的方法在保证机器人运动模仿的稳定性前提下很好地提高了行为模仿的速度。The imitation learning method of robot is restricted by the speed of demonstrator,which limits the speed of the robot’s imitation,and cannot give full play to the performance of the robot.To improve the rapidity of robot behavioral imitation,a method for adaptively changing the speed of robot imitation learning movement is proposed.Firstly,the demonstration movement is modeled by the method based on dynamical system,and the sufficient condition of dynamical system stability is taken as a constraint to ensure the stability of behavioral imitation.Secondly,a nonlinear function which varies with the distance from the current state of the robot to the target point is constructed.The nonlinear function is used as a parameter to combine with the system model to adaptive adjust the speed of behavioral imitation.Finally,four methods of imitation learning evaluation are given to evaluate the performance of imitation.The experimental results show that the proposed method improves the speed of behavioral imitation under the premise of maintaining the stability of robot behavioral imitation.

关 键 词：机器人 模仿学习 动态系统 非线性函数 性能评价 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]