Development and Evaluation of Positioning Systems for Autonomous Vehicle Navigation

摘要

Successful navigation of an autonomous vehicle is made possible by accurately measuring real-time position and orientation. Positioning systems should provide fast, accurate and reliable operation through varying conditions. The current positioning system being used on Center for Intelligent Machines And Robotics (CIMAR) Navigation Test Vehicle (NTV) consists of an Inertial Navigation System (INS) and Global Positioning System (GPS) integrated through an external Kalman Filter (KF). The INS component is a Honeywell H-726 Modular Azimuth Positioning System (MAPS) and the GPS component is an Ashtech Z- 12 Differential GPS. The high data rate of the INS directly complements the high accuracy of the GPS. System performance shows sustained positional accuracy of less than ten centimeters at output data rates of up to 10 Hz. The high system cost, however, limits its versatility in application. The focus of this thesis is to explore two alternative low-cost positioning systems. A second positioning system is the Novatel Beeline GPS, which consists of dual antennas to measure both vehicle position and dual-axis orientation. The drawbacks of the system are inherent to GPS, including satellite visibility at all times and slow recovery after data loss. This stand-alone system features position data to within 20 centimeters at data rates up to 5 Hz. A third system is a low-cost INS/GPS integrated positioning system. It makes use of a Honeywell HG1700AG11 Inertial Measurement Unit (IMU) and a Novatel RT-20 Differential GPS. Raw acceleration and angular velocity data at 100 Hz from the IMU is averaged down to 12.5 Hz for processing by the Navigation Processor. The Processor performs a two-stage alignment to determine the IMU's initial orientation angles. The second alignment stage uses a Kalman Filter to reduce platform tilt errors. Once aligned, data is used to plot out a navigation solution to provide position, velocity and orientation.