Intelligent Vehicle Collision Avoidance System Based On LIDAR

Vehicle active safety technology has been a hot area of research in almost all the vehicle manufacturers and research institutes all over the world. To make the vehicle safer and smarter is the developing direction of the future automotive industry. The intelligent vehicle collision avoidance system based on LIDAR is one of the most important research branches of the active safety technology. The main research contents are as follows:(1) The design of LIDAR point cloud collecting hardware platform: this paper contains the work of an independent design of a LIDAR point cloud collecting hardware platform. This platform is based on a Freescale MC9S08DZ60 microcontroller and it has been proved to be able to collect and decode the special data type of SICK LMS151 LIDAR. Meanwhile, the platform is also capable of analyzing the data decoding result, sending alerts to the driver automatically and uploading the data to an upper computer remotely.(2) A proposal of an optimized DBSCAN clustering algorithm: this paper proposed an optimized DBSCAN clustering algorithm. The traditional DBSCAN clustering algorithm will be faced with a series of problems including the difficulties to choose proper clustering parameters when the data space density comes to uneven. LIDAR point cloud data is a typical data type whose data space density is not uniform, to be specific, data points those who are further from the LIDAR have a lower density compared with those who are closer to the LIDAR. Aiming at these properties, the paper proposes a new algorithm which added the distance factor to the clustering parameter of the traditional DBSCAN clustering algorithm to make it more adaptive to the LIDAR data characteristics.(3) Experiments aiming at the hardware platform and the clustering algorithm: the paper contains a large amount of experiments to verify the capability and functional completeness of the hardware platform. It also contains experiments to compare the difference clustering effects between the traditional DBSCAN algorithm and the optimized DBSCAN algorithm. These experiments have eventually proved the effectiveness and superiorities of the new algorithm.