基于多传感器信息系统的仿人机器人步态规划算法改进研究

Development of Robot Gait Planning Algorithm Based on Multi-Sensor Information System

摘要：基于视觉传感器、角度传感器和力／力矩传感器组成的多传感器信息系统，对仿人机器人理想的步态规划算法进行了改进．以双目视觉立体标定原理处理视觉传感信息，从而判断目标距离及目标路径的可达性；通过角度传感器获取实时旋转角度，经参数调整减小执行误差；根据力／；5矩传感信息对支撑脚踝关节的侧摆角度进行增量式补偿．经实验验证，改进后的算法对机器人的步态稳定性控制具有良好的效果．

The algorithm for ideal robot gait planning is improved based on the multi-sensor information system consisting of vision sensors, angle transducers and force/torque sensors. The information of vision sensors is dealt with stereo binocular vision calibration theory so as to estimate the distance of the destination and whether the destination is reachable or not. Real-time rotation angle is obtained through angle transducers so as to reduce execution error by adjusting parameters. Incremental compensation to scroll angle of supporting foot ankle is carried out according to the information of force/torque sensors. The experiment results show that the improved algorithm has good controlling effect on the stability of the robot＇s gait.