机器人立体视觉模块的故障诊断

Fault diagnosis of stereo vision module for the robot

针对机器人的视觉模块,提出了一种故障诊断的算法.该方法根据极线几何理论,利用基本矩阵离线地得到故障诊断的期望范围,再根据匹配像素点是否落在期望范围内判断视觉模块当前是否存在故障.与常规方法相比,不需要增加额外的传感器,且占用系统资源较少,具有相当好的实时性.实验证明,利用该方法进行故障诊断的错判率和漏检率均在3%以下.

An algorithm of fault diagnosis was proposed for vision module on the robot. The expected area was generated based on the fundamental matrix, an important concept of epipolar geometry. The diagnosis can be performed in real time by judging whether matched pixels in the expected area. Compared with the conventional methods, the method doesn′t need additional sensors and requires less CPU load. The experiments showed that the error rates of false-positive and false-negative judgments are below 3%.