张亚徽, 王斐, 李景宏, 刘玉强, 吴仕超. 基于稳态视觉诱发电位的智能轮椅半自主导航控制[J]. 机器人, 2019, 41(5): 620-627,636. DOI: 10.13973/j.cnki.robot.180771
引用本文: 张亚徽, 王斐, 李景宏, 刘玉强, 吴仕超. 基于稳态视觉诱发电位的智能轮椅半自主导航控制[J]. 机器人, 2019, 41(5): 620-627,636. DOI: 10.13973/j.cnki.robot.180771
ZHANG Yahui, WANG Fei, LI Jinghong, LIU Yuqiang, WU Shichao. Semi-autonomous Navigation Control of Intelligent Wheelchair Based on Steady State Visual Evoked Potential[J]. ROBOT, 2019, 41(5): 620-627,636. DOI: 10.13973/j.cnki.robot.180771
Citation: ZHANG Yahui, WANG Fei, LI Jinghong, LIU Yuqiang, WU Shichao. Semi-autonomous Navigation Control of Intelligent Wheelchair Based on Steady State Visual Evoked Potential[J]. ROBOT, 2019, 41(5): 620-627,636. DOI: 10.13973/j.cnki.robot.180771

基于稳态视觉诱发电位的智能轮椅半自主导航控制

Semi-autonomous Navigation Control of Intelligent Wheelchair Based on Steady State Visual Evoked Potential

  • 摘要: 针对现有基于脑-机接口(BCI)控制的智能轮椅因交互不协调、识别准确率低、执行效率差而造成用户疲劳等问题,提出了一种人机协同智能控制方法,设计并实现了一种基于BCI与层级地图结合的智能轮椅半自主导航控制系统.首先根据实际需要构建栅格-拓扑-意图3级层级地图.然后采用基于典型相关性分析的1维卷积神经网络对人的意图进行识别分类,并通过BCI系统发送至导航控制部分.最后经融合决策给出控制指令实现智能轮椅的导航.实验中所提方法的平均准确率为91.576%,对脑电信号的识别准确率高,控制系统的稳定性好.结果表明该方法可灵活控制轮椅运动方向并按照人的控制意图无碰撞地到达目标地点.

     

    Abstract: Current brain-computer interface (BCI) control based intelligent wheelchairs are facing the problem of users' fatigue caused by uncoordinated interaction, low recognition accuracy, low execution efficiency. A method of human-machine collaborative intelligent control is proposed to solve the problem, and a semi-autonomous navigation control system based on BCI and hierarchical map is designed and implemented for intelligent wheelchair. Firstly, a three-level raster-topology-intention map is constructed according to actual needs. Then, a one dimensional convolutional neural network (1D-CNN) based on canonical correlation analysis (CCA) is used to identify and classify the human's intention, which is sent to the navigation control section through BCI system. Finally, the control command is given to control the navigation of the intelligent wheelchair by fusion decision. The experimental results show that the proposed method achieves high recognition accuracies on electroencephalography (EEG) signals with an average of 91.576%, and the control system demonstrates a good stability. The method can flexibly control the motion direction of intelligent wheelchair and reach the target location without collision according to the human's control intention.

     

/

返回文章
返回