UC Riverside

Traditional positioning solely using GPS has been shown to be inadequate for more advanced Intelligent Transportation System (ITS) applications. Accurate and reliable mapping also requires a highly accurate and reliable positioning system. Integration of GPS and an Inertial Measurement Unit (IMU) is a promising method in terms of improving accuracy. When GPS signal are unavailable to reset IMU errors, range sensors, e.g., RADAR and LiDAR which can measure distance and angle to road side landmarks by actively emitting power and measuring reflected signals, serve as a good complement to guarantee accuracy. In addition, the advancement in 3D LiDAR technology makes the mobile mapping system very handy because it does not require lane closures or time-consuming human surveying, thereby saving both time and money.

In this dissertation, a novel automotive RADAR-aided Differential GPS/INS system is presented. The RADAR measurement model is analyzed, and proper types of landmarks are investigated and verified. The residue and corresponding error models are also analyzed. Two separate mathematical models are proposed for integration with GPS/INS in an Extended Kalman Filter (EKF) architecture. Experiments in a controlled environment are described and the results illustrate significant improvement of positioning accuracy when the RADAR detects landmarks, data association is successful, and RADAR measurements are used to update the EKF estimates.

In the second part of the dissertation, a 3D LiDAR-based Mobile Mapping system is presented. The overall system architecture on both hardware and software are demonstrated. An intersection stop bar extraction algorithm based on image processing is then described in detail with intermediate results demonstrated as images. The results of the algorithm are the accurate position of the endpoints of each stop bar in a global coordinate frame.