节点文献
助力机器人传感器信号倍频算法与髋关节并联机构控制系统研究
Research on Algorithm for Multiplying the Frequency of the Sensor Signal and Control System for the Parallel Mechanism on Hip Joint
【作者】 姚俊章;
【导师】 余永;
【作者基本信息】 中国科学技术大学 , 检测技术与自动化装置, 2011, 硕士
【摘要】 可穿戴助力机器人是近年来新兴起来的研究领域,它涉及人体解剖学、人体工程学、运动生物力学、机器人学、机构学、人工智能以及信息融合等学科。可穿戴助力机器人是一种典型的用于扩展人类运动能力的外骨骼助力装置。它能够根据感知系统获得的人体运动信息进行自主控制,从而对人体的运动提供助力。感知系统对人体运动信息的获取是否及时准确将会影响到可穿戴机器人的助力效果。为了改善传感器信号的滞后性和响应频率不够高的现象,本文提出了一种基于实时预测的传感器信号倍频算法。该算法综合了预测算法的提早预见性,以及数值插值算法能够对离散信号的光滑补偿性的优点,不仅可以实时地提早获得人体运动意图,还可以成倍地提高传感器信号的响应频率。同时针对串联机构不能很好地符合人体髋关节的生理特点及其机械特性上的不足,本文提出了利用并联机构来设计人体下肢髋关节处的助力装置,并在对该并联机构搭建了控制系统,本文通过对并联机构前屈后伸自由度进行运动学分析后,提出了基于假想柔顺控制策略与逆雅克比关系式的髋关节并联机构控制决策,为以后深入研究并联机构助力打下基础。本文具体内容如下:一、从信号角度出发,本文提出一种基于滑动窗口二次自回归(MWQAR)模型和误差修正的实时预测算法。该算法由时间序列模型滑动窗口二次自回归(MWQAR)模型及其相关参数的估计算法和基于预测误差的预测值补偿方法共同构成。传感器信号预测实验结果表明该预测算法能够提早感知人体运动的信息。二、在实时预测算法的基础上,本文提出了一种传感器信号倍频算法,用以提高信号的响应频率。该算法由基于滑动窗口二次自回归(MWQAR)模型和误差修正的实时预测算法和三次样条插值法构成。文章通过对传感器信号进行倍频算法实验,进一步说明该算法是可行且有效的。三、本文为基于并联机构的髋关节助力装置设计了控制系统。在分析了基于人机接触力信息的假想柔顺控制策略以及其自适应调整后,本文针对并联机构单个自由度上的运动学进行了分析,并建立了机构的主动关节转动速度与机构在该自由度方向上的转动速度间的逆雅克比关系式。本文设计的髋关节助力装置控制系统的决策原理就是基于假想柔顺控制策略以及此逆雅克比关系式建立的。本文还利用基于该原理的控制软件完成了人体髋关节前屈后伸运动中助力装置的控制实验。
【Abstract】 In recent years, the research on the Wearable Power Assist Robot become more popular, which involves a lot of disciplines, such as human anatomy, ergonomics, biomechanics, robotics, mechanism, artificial intelligence and information fusion. The Wearable Power Assist Robot is a typical exoskeleton, which can expand human’s motion capacities. The robot can drive itself according to motion information of the human obtained by the sensor system, so that it can provide power to assist human’s motion.Whether the information about the human’s motion obtained by the sensor system is timely and accurate or not may affect the effect of the power assist robot. In order to improve the delay and the lower response frequency of the sensor signal, we present an algorithm to multiply the sensor signal’s response frequency based on real-time predicting. This algorithm has the advantages about the predictability and sleek compensatory. Therefore, it not only can gain the human’s motion information earlier, but also can increase the sensor signal’s frequency many times.Because of the lack of the serial mechanism’s mechanical properties and the physiological characteristics of the human hip joint, we decide to use parallel mechanism to design Power Assist Robot for human hip joint, and design the control system for it. Based on analyzing the kinematics of the parallel mechanism on only one degree of freedom what is in the direction of flexion and extension, we propose a method of decision of the control system based on the virtual compliance control method and the inverse Jacobian matrix on only one degree of freedom, which is the basis for the further research on Parallel Mechanism for power assist.This paper’s main contents are as follows:1. From the view of the signal, we find an algorithm for real-time predicting based on the Moving Windows Quadratic Autoregressive (MWQAR) model and the improvement of the error. The algorithm contains the method of modeling the signal by the MWQAR model and estimating its parameters, and the method of using forecasted error to improve the effect of the predicting algorithm. The results of the experiments show that the algorithm can advance gaining the information of the human motion.2. Based on the real-time predicting algorithm, we propose a method to multiply the sensor signal’s response frequency. The method is composed by the real-time predicting algorithm and the cubic spline interpolation method. Through the results of using the multiplying algorithm on the sensor signal, we can find that the algorithm is feasible and effective.3. We design a control system for the Power Assist Robot for the hip joint based on the parallel mechanism. While analyzing the virtual compliance control strategy based on interaction force and its adaptive adjustment, we analysis the kinematics of the parallel mechanism on only one degree of freedom, and find its Jacobian matrix between the speed of the mechanism’s active joint and the speed of the hip joint in this direction. Based on the virtual compliance control strategy and the Jacobian matrix, we design the decision system of the Power Assist Robot’s control system. By applying the software of the control system based on the decision system, we finish the experiments of controlling the Power Assist Robot to help the hip joint to move in flexion and extension.
【Key words】 Power Assist Robot; real-time predicting; multiplying the sensor’s frequency; parallel mechanism; virtual compliance control;