Li Wei, Li Mengfan, Zhao Jing
Department of Computer and Electrical Engineering and Computer Science, California State University Bakersfield, CA, USA ; School of Electrical Engineering and Automation, Tianjin University Tianjin, China.
School of Electrical Engineering and Automation, Tianjin University Tianjin, China.
Front Syst Neurosci. 2015 Jan 9;8:247. doi: 10.3389/fnsys.2014.00247. eCollection 2014.
This paper investigates controlling humanoid robot behavior via motion-onset specific N200 potentials. In this study, N200 potentials are induced by moving a blue bar through robot images intuitively representing robot behaviors to be controlled with mind. We present the individual impact of each subject on N200 potentials and discuss how to deal with individuality to obtain a high accuracy. The study results document the off-line average accuracy of 93% for hitting targets across over five subjects, so we use this major component of the motion-onset visual evoked potential (mVEP) to code people's mental activities and to perform two types of on-line operation tasks: navigating a humanoid robot in an office environment with an obstacle and picking-up an object. We discuss the factors that affect the on-line control success rate and the total time for completing an on-line operation task.
本文研究通过运动起始特定的N200电位来控制人形机器人的行为。在本研究中，通过在直观代表要用思维控制的机器人行为的机器人图像上移动一条蓝色条带来诱发N200电位。我们展示了每个受试者对N200电位的个体影响，并讨论了如何处理个体差异以获得高精度。研究结果表明，超过五名受试者击中目标的离线平均准确率为93%，因此我们使用运动起始视觉诱发电位（mVEP）的这一主要成分来编码人们的心理活动，并执行两种在线操作任务：在有障碍物的办公室环境中导航人形机器人以及拾取物体。我们讨论了影响在线控制成功率和完成在线操作任务总时间的因素。
