2016 年 3 巻 1 号 p. 14-00546
The total length of sewer pipelines in Japan is over 440,000 kilometers. Until now, experienced inspectors have conducted visual inspections by watching in-pipe animation images that are obtained from an autonomous vehicle equipped with a CCD camera. Their inspection methods have difficulties of quantitative evaluation with high accuracy because the viewpoints depend on the judging skills of the inspectors. In this study, we developed vehicles equipped with an inertial navigation system (INS) consisting of a gyro sensor and accelerometer in order to measure unevenness of the sewer pipe accurately and quantitatively. We also aim to achieve a simple measurement scheme at comparatively low cost. To accomplish these missions, two vehicles were designed and manufactured. The first vehicle is designed for the function of driving and pulls a second vehicle. The second vehicle’s only function is measurement; it is towed and does not move independently. We use the MEMS sensor devices installed on the second vehicle to suit our particular needs of low price and small size. However, these low-priced gyro sensors are notorious for their inaccuracy (experiencing such problems as drift error). We apply the extended Kalman filter (EKF) algorithm to reduce the estimation errors. We use quaternions as state variables of the posture for the 3D coordinate system. The experimental study shows that the suggested algorithms effectively remove the errors and can lead to systematic measurement schemes to accurately determine the unevenness of the sewer pipes.
