碩士 / 國立臺灣大學 / 機械工程學研究所 / 106 / This research aims to propose an autonomous vehicle localization and navigation system for functional safety requirements (FSR) of electrical/electronic systems in road vehicles. The main functional safety concept (FSC) includes such as fault detection, fail-safe, and fault tolerance. For the issue, this research uses a real-time multi-sensor fusion localization technique. When one of sensors fails, this method can extend the reliable positioning time for decision system to do the right decision. This approach is based on the map-guided technique and unscented Kalman filter with Global Navigation Satellite System (GNSS) and Inertial Navigation System (INS).However, the sigma points which are calculated from unscented transform might be located in unreasonable place, such as sidewalk. Using the detection data from LiDAR to be the constraint can restricted the sigma points which are out of constraint in the reasonable region. To design the constrained unscented Kalman filter, this research combines the detection data from LiDAR and digital map to determine the range of constraint. This method can not only reduce positioning estimation error, but also detect the positioning state for control system that can reduce danger by wrong information. Finally, this research uses an electric medium-sized bus as a verification platform to prove the localization system can run safe in different scenarios by human driving and autonomous driving.
Identifer | oai:union.ndltd.org:TW/106NTU05489140 |
Date | January 2018 |
Creators | Hsiang-Wen Hou, 侯翔文 |
Contributors | Kang Li, 李綱 |
Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
Language | zh-TW |
Detected Language | English |
Type | 學位論文 ; thesis |
Format | 157 |
Page generated in 0.0099 seconds