基于数据驱动的膝关节外骨骼控制

为了识别人体运动意图协调人机运动，采用二维激光测距仪采集地形数据进行在线识别，使用学习向量量化（LVQ）的方法，基于不同地形间的距离特征实现快速、准确的地形分类. 设计基于数据驱动的无模型自适应控制方法，基于膝关节角度的输入输出数据建立动态线性化模型，避免了人机外骨骼建模的复杂性和建模误差. 建立人机外骨骼模型，通过仿真得到正常行走时膝关节的先验力矩，引入先验力矩提高控制器的准确性. 搭建ADAMS和MATLAB联合仿真平台，选取平地路况进行实验. 实验结果表明，所设计的控制方法使得外骨骼膝关节对目标角度有良好的跟踪，对人体行走有较好的助行效果.

Knee-joint exoskeleton control based on data-driven approach

The two-dimensional laser rangefinder was used to collect terrain data for online identification in order to identify human movement intentions and coordinate human-exoskeleton motion. The method of learning vector quantization (LVQ) was used based on the distance features between different terrains in order to achieve fast and accurate terrain classification. A model-free adaptive control method based on data drive was designed, and the dynamic linearization model was established based on the input and output data of knee joint angle, which avoided the complexity and error of human-exoskeleton modeling. A human-exoskeleton model was established and the prior torque of the knee joint was obtained through the walking simulation. The prior torque was introduced to improve the accuracy of the controller. The ADAMS-MATLAB co-simulation platform was constructed, and the flat road condition was selected for experiment. The experimental results show that the designed strategy enables the knee-joint exoskeleton to track the trajectory of angle well and has a good performance on walking assistance.