Vehicle Automation System Based on Multi-sensor Integration

The purpose of the study is to develop the guidance system by the sensor fusion integration with a machine vision, an RTKGPS and a geometric direction sensor (GDS). The Extended Kalman Filter (EKF) and a statistical method based on twodimensional Probability Density Function are adopted as a fusion integration methodology in this research. To achieve the navigation planner based on sensor fusion integration, the four types of control strategies were built by changing combination of three kinds of sensors; machine vision, the RTK-GPS and the GDS. The developed navigation planner involves the priority scheme of the control strategies by knowledge-base approach. The average lateral error of the vehicle guidance based on the fusion of the RTK-GPS and the GDS indicated 8.4 cm. Because the lateral error was less than 20 cm which was the accuracy of the RTK-GPS, the developed sensor fusion methodology with the EKF was seemed to be satisfactory operation.