Abstract
Keywords:
Autonomous Driving, DGPS, RTK
Abstract
Simulation study for the development of a DGPS RTK-based autonomous driving system was carried out using a dSpace Micro-AutoBox as the autonomous driving computer and a 3D vehicle model. The positioning error model of the vehicle in motion used in the simulation is based on the result of the DGPS RTK driving test. The error is composed of white noise, first-order Markov process and a bias term due to the latency of the DGPS system. For the driving test, NovAtel DL-4 GPS receivers, which provide 3cm CEP (95%) positioning accuracy, were used and an additional external system was set up to measure the positioning accuracy of the vehicle. For the control of the autonomous driving system, an accurate digital map of the test ground is required to provide the reference line for the autonomous vehicle to follow. Smooth curve data for the test lane was generated by polynomial curve fitting with repeatedly collected position data of the lane using the DGPS RTK. The longitudinal and lateral control of the autonomous driving system were considered independently, and the curvature information of the test lane and the latency due to data processing and communication were taken into account in computing the control commands to show the performance improvement of the autonomous driving system.