Application of Homography Transformation for Assisting Machine Vision in Automatic Guided Vehicles

The safety of autonomous car driving is an important concern raised in the recent years. With the development of various sensors, lots of driving assistant device are applied in the vehicle, such as sonar sensing and inertial navigation Other than terms listed above, machine vision is also a momentous technology in the automatic guided vehicle. Automatic guided vehicle integrates technologies such as machine vision and automatic control, and expects to use computer-assisted driving to enhance the safety of driving. However, the location of the camera installation limited by car structures could cause the camera cannot properly record the road situation at the optimal observation point. Consequently, the image could be significantly distorted. Therefore, the image recorded by camera have to be corrected in order to reduce the image distortion. This study utilizes a webcam to obtain image of the surrounding, and process the image with LabVIEW for promote the precision of geometric matching via homography transformation. Meanwhile, for testing the effectiveness of the method, an automatic guided vehicle is also realized by hiring IG-42 motors and using a MyRIO board and LabVIEW as the program control and the central information hub. After integrating the perspective calibration for geometric matching and automatic guided vehicle , we conducted an experiment of tracking a mark line, which is commonly used in manufacturing factories. In addition, we also tested the geometric matching of the image after correction. The distinguish rates rose from 0.63 to 0.93, represents uncorrected and corrected respectively. Finally, we finish an automatic guided vehicle by machine vision. A more complete vision automatic guided vehicle system will go on in the future.