室内机器人动态手势指令识别系统研究

摘要

针对目前室内移动机器人手势指令识别系统存在的问题,对图像传感器与机器人相分离的图像采集方案进行了研究,并利用动态手势指令对机器人进行控制.动态手势指令识别方法是对手的不同运动轨迹进行识别,通过皮肤颜色模型和手势中心点方向向量法追踪得到手势运动轨迹,提取手势运动轨迹的特征向量,通过基于动态时间规整(DTW)实现对轨迹的识别.实验结果表明,该系统可以实现对机器人前进、后退、左转、右转的实时控制.%A novel method of gesture instruction recognition system for indoor mobile robot is proposed by separating the image sensor and the robot, and controlling the robot by dynamic gesture instruction recognition, which is used to identify the different trajectories.The trajectory of the gesture is gotten by the skin color model combining the direction vector of the gesture's center-point.The feature vector of the gesture trajectory is extracted.DTW is used to identify the trajectory.Ex-perimental results demonstrate that this system realizes the real-time controlling of forward, backward, turning left, turning right to indoor robot.