矢量场逐次逼近的康复机器人柔顺交互控制

为了实现康复机器人的主动柔顺交互，提出了一种基于矢量场逐次逼近的控制模型；设计了矢量场逐次逼近系统，可输出机器人关节期望位移，该输出能与输入的扭矩、表面肌电及脑电等信号在振幅、频率和相位上保持同步，且通过调节遗忘因子参数值，可改变主动柔顺交互的积极性；利用自行设计的穿着型下肢康复机器人样机进行柔顺辅助实验，以验证所提出控制模型的有效性；通过FFT（Fast Fourier transformation）频谱对机器人关节扭矩的组成成分进行了分析，并采用基于最小二乘法的参数辨识方法实施了重力补偿，以便康复机器人实时控制.实验结果表明，该控制模型对于实现康复机器人与人之间的柔顺交互是有效的.

Rehabilitation Robot Compliance Interaction Control With Successive Approximation Vector Field

In order to realize active compliant interaction for rehabilitation robots, we propose a control model based on a successive approximation method of vector field. A system in the vector field with successive approximation is designed to output the desired trajectory of each robot joint. The amplitude, frequency and phase of the output can synchronize with the torque, surface electromyogram and electroencephalogram information of the input signal; the activity of compliant interaction can be changed by adjusting a forgetting factor in the system. Based on a self-designed wearing type lower limb rehabilitation robot prototype, a compliance assistant experiment is presented by adopting the proposed control model. The components of the joint torque for the robot are analyzed using the fast Fourier transformation (FFT) spectrum analysis method, and gravity compensation for the rehabilitation robot joints is realized for real-time control by using parameter identification based least square method. Experiment results indicate that the control method is effective for the compliant interaction between rehabilitation robot and human.