隨著自動化的蓬勃發展，自動駕駛近年來成為一個熱門的研究領域。在這門研究領域中，自動駕駛車輛的環境感知能力、決策分析及追跡系統固然重要，然而通常眾人最在意的實際議題為自駕車之安全性。一般的自動駕駛車輛皆搭載多項感測器來提高環境感知能力以達到良好的定位效果，例如光達(LiDAR)、雷達(radar)、相機、全球導航衛星系統（Global Navigation Satellite System, GNSS）接收機、慣性測量單元（Inertial Measurement Unit, IMU）等皆可用來精進車輛導航定位系統，以得知車輛精確位置而不會行駛出車道邊緣。但在高爾夫球場中，地勢起伏變化大、車道兩旁的高聳樹木、多處的隧道等因素皆會影響感測器之效能，導致車輛定位不穩定，危害乘客安全。
本文提出了一種無論定位是否精確皆適用之即時自動駕駛道路邊界偵測及追蹤系統，適用於高爾夫球場中的地勢起伏變化大、非結構道路情境。先利用相對低成本混合式固態光達進行道路邊緣偵測，再使用自動駕駛車輛定位解及慣性量測單元搭配卡爾曼濾波器(Kalman filter)來追蹤道路邊緣。另外，也在事前建置完成的點雲地圖(Point Cloud Map)中繪製高爾夫球場的車道邊緣線，視為高精地圖(High Definition Maps, HD Maps)，當作另一種道路邊緣辨識系統。最後使用高階資訊融合(High-level Fusion)將兩種道路邊緣偵測結果融合，並且在定位不精確等不安全情況時使自動駕駛車輛減速至停止，讓駕駛人員接手操控車輛。

With the vigorous development of automation, autonomous driving has become a popular research field in recent years. In this research field, while the environmental perception capability, decision making, and tracking system of autonomous vehicles are important, the safety of self-driving cars is generally the most concerning practical issue. Generally, autonomous vehicles are equipped with multiple sensors to enhance environmental perception capability for more accurate positioning analysis. Sensors such as LiDAR, radar, cameras, Global Navigation Satellite System (GNSS) receivers, and Inertial Measurement Units (IMUs) can be employed to improve the navigation and positioning performance of the vehicle, ensuring it stays within the road boundaries without deviation. However, on a golf course, factors such as changes in terrain, towering trees along the road boundary, and tunnels can affect the performance of the sensors, leading to inaccurate vehicle positioning solution and endangering passenger safety.
This thesis proposes a real-time system for road boundary detection and tracking regardless of the accuracy of positioning. It is designed for the rugged and non-structured road on a golf course. The system utilizes a relatively low-cost hybrid solid-state LiDAR for road boundary detection and then tracks the road boundary with Kalman filter using autonomous vehicle positioning solution and inertial measurement. Additionally, another road boundary detection system uses a previously generated Point Cloud Map to draw road boundaries to serve as a High Definition Map (HD Map). Finally, the two road boundary detection results are fused using high-level fusion. Adopting the proposed method, in dangerous situations such as vehicle positioning is inaccurate, the autonomous vehicle will decelerate to a stop, allowing the driver to take control of the vehicle.

[1] R. E. Kalman, “A New Approach to Linear Filtering and Prediction Problems,” Transaction of the ASME - Journal of Basic Engineering, pp. 35-45, 1960.
[2] A. Seibert, M. Hähnel, A. Tewes and R. Rojas, "Camera based detection and classification of soft shoulders curbs and guardrails", in Proceedings of the IEEE Intelligent Vehicles Symposium (IV), pp. 853-858, Jun. 2013.
[3] M. B. de Paula and C. R. Jung, "Automatic detection and classification of road lane markings using onboard vehicular cameras", IEEE Transactions on Intelligent Transportation Systems, vol. 16, no. 6, pp. 3160-3169, Dec. 2015.
[4] C. Fernandez, D. Fernandez-Llorca and M. A. Sotelo, "A hybrid vision-map method for urban road detection", Journal of Advanced Transportation, vol. 2017, Sep. 2017.
[5] D. Sodhi, S. Upadhyay, D. Bhatt, K. M. Krishna and S. Swarup, "CRF based method for curb detection using semantic cues and stereo depth", in Proceedings of 10th Indian Conference on Computer Vision, Graphics and Image Processing, pp. 41, 2016.
[6] Z. Wang, G. Cheng and J. Y. Zheng, "All weather road edge identification based on driving video mining", in Proceedings of IEEE 20th IEEE Intelligent Transportation Systems Conference (ITSC), pp. 1-6, Oct. 2017.
[7] S. Kakegawa, H. Matono, H. Kido and T. Shima, "Road surface segmentation based on vertically local disparity histogram for stereo camera", International Journal of Intelligent Transportation Systems Research, vol. 16, no. 2, pp. 90-97, 2018.
[8] B. Jia, J. Chen and K. Zhang, "Drivable road reconstruction for intelligent vehicles based on two-view geometry", IEEE Transactions on Industrial Electronics, vol. 64, no. 5, pp. 3696-3706, May 2017.
[9] W. S. Wijesoma, K. R. S. Kodagoda and A. P. Balasuriya, "Road-boundary detection and tracking using ladar sensing", IEEE Transactions on Robotics and Automation, vol. 20, no. 3, pp. 456-464, Jun. 2004.
[10] R. Huang, J. Chen, J. Liu, L. Liu, B. Yu and Y. Wu, "A practical point cloud based road curb detection method for autonomous vehicle", Information, vol. 8, no. 3, pp. 93, 2017.
[11] D. Pfeiffer and U. Franke, "Efficient representation of traffic scenes by means of dynamic stixels", in Proceedings of IEEE Intelligent Vehicles Symposium, pp. 217-224, Jun. 2010.
[12] Y. Zhang, J. Wang, X. Wang and J. M. Dolan, "Road-Segmentation-Based Curb Detection Method for Self-Driving via a 3D-LiDAR Sensor", IEEE Transactions on Intelligent Transportation Systems, vol. 19, no. 12, pp. 3981-3991, Dec. 2018.
[13] B. Yang, L. Fang and J. Li, "Semi-automated extraction and delineation of 3D roads of street scene from mobile laser scanning point clouds", ISPRS Journal of Photogrammetry & Remote Sensing, vol. 79, pp. 80-93, May 2013
[14] J. H. Song, S. H. Han, K. Y. Yu and Y. I. Kim, "Assessing the possibility of land-cover classification using LiDAR intensity data", International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, vol. 34, pp. 259-262, 2002.
[15] M. A. Fischler and R. C. Bolles, "Random sample consensus: A paradigm for model fitting with applications to image analysis and automated cartography", Communications of the ACM, vol. 24, no. 6, pp. 381-395, 1981.
[16] J. D. Gergonne, "The application of the method of least squares to the interpolation of sequences", Historia Mathematica, vol. 1, pp. 439-447, Nov. 1815.
[17] M. Himmelsbach, F. V. Hundelshausen and H.-J. Wuensche, "Fast segmentation of 3D point clouds for ground vehicles", in Proceedings of IEEE Intelligent Vehicles Symposium, pp. 560-565, Jun. 2010.
[18] “SAE international”, https://www.sae.org/ (accessed Jun. 15, 2023)
[19] “Autoware Documentation”, https://autowarefoundation.github.io/autoware-documentation/main/design/autoware-architecture/ (accessed Jun. 15, 2023)
[20] “ROS-Robot Operating System”, https://www.ros.org/ (accessed Jun. 15, 2023)
[21] M.Quigley et al., “ROS: an Open-Source Robot Operating System,” in ICRA Workshop on Open Source Software, vol. 3, no. 3.2, 2009.
[22] “RQT | ROS tool”, http://wiki.ros.org/rqt (accessed Jun. 15, 2023)
[23] “RViz | 3D visualization tool for ROS”, http://wiki.ros.org/rviz (accessed Jun. 15, 2023)
[24] H. Lim, M. Oh and H. Myung, "Patchwork: concentric zone-based region-wise ground segmentation with ground likelihood estimation using a 3D LiDAR sensor", IEEE Robotics and Automation Letters, vol. 6, no. 4, pp. 6458-6465, 2021.
[25] “Image Segmentation Keras : Implementation of Segnet, FCN, UNet, PSPNet and other models in Keras”, https://github.com/divamgupta/image-segmentation-keras (accessed Jul. 10, 2023)
[26] “Keras: Deep Learning for humans”, https://keras.io/ (accessed Jul. 10, 2023)
[27] K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition”, in arXiv:1409.1556, 2014
[28] “Classic Platform - AUTOSAR”, https://www.autosar.org/standards (accessed Jul. 10, 2023)
[29] “Vector Map Builder”, https://tools.tier4.jp/vector_map_builder_ll2/ (accessed Jul. 10, 2023)
[30] “Neousys Technology: Nuvo-5095GC”, https://www.neousys-tech.com/en/product/ product-lines/edge-ai-gpu-computing/nuvo-5095gc-gpu-computer (accessed Jul. 25, 2023)
[31] “Robosense LiDAR M1”, https://www.robosense.ai/en/rslidar/RS-LiDAR-M1 (accessed Jul. 25, 2023)
[32] “Intel RealSense Depth Camera D455”, https://www.intelrealsense.com/depth-camera -d455 (accessed Jul. 25, 2023)
[33] “u-blox C100-F9K application board”, https://www.u-blox.com/en/product/c100-f9k-application-board (accessed Jul. 25, 2023)
[34] “E-Z-GO Freedom RXV golf cart”, https://ezgo.txtsv.com/sites/default/files/2020-10/Freedom%20RXV.pdf (accessed Aug. 15, 2023)