Integrated Camera/Laser/IMU/RTK-GPS Localization System for Autonomous Vehicles

  • 游 尚霖

Student thesis: Master's Thesis


Autonomous vehicle is getting more and more attention today With autonomous vehicle people get more free time during driving Traffic jams can be reduced and traffic accidents caused by human errors may be lower A key technology for an autonomous car is the ability to localize itself accurately For accurate positioning one method is to localize the car by prebuilt maps Navigation systems match the measurements from external sensors like laser scanners or cameras to the map and determine the position of the vehicle However building dense maps is time consuming and complex and systems need a mass storage device for the maps To ease the relying on dense maps real time kinematic satellite navigation system is a way to fulfill the requirement but the performance of satellite navigation systems may degrade due to multipath effect or signal attenuation Hence navigation systems must diagnose the abnormal GPS results and employ different odometry techniques for the navigation tasks under GPS-denied environments Besides all sensors are subject to errors It is critical for systems to decide which and when the sensor measurements are reliable The thesis studies the use of RTK-GPS based navigation system to navigate the autonomous car without prebuilt dense maps Real world experiments show the capability of the system to navigate the driverless car successfully under normal condition To account the navigation tasks under GPS-denied environments odometry techniques which can be used to back up the degradation or loss of the RTK-GPS signal are investigated Different odometry techniques include point cloud scan matching visual odometry and inertial navigation systems are described For sensor integrity scenarios lead to performance degradation of odometry techniques are investigated and analysis of the uncertainty of the odometry techniques is studied Finally real world experiments are carried out to evaluate the sensor fusion navigation system
Date of Award2017 Mar 21
Original languageEnglish
SupervisorJyh-Chin Juang (Supervisor)

Cite this