Radar Algorithm Research For Avoidance Of Secondary Collision At The Scene Of Highway Accident

Since twenty-first century, development of intelligence transportation has brought about multifold influence for the entire road traffic system and it has offered better conditions and services for everyone's daily life, traffic control, and public safety. The commercial microwave radars which serve as primary sensors utilized in the ITS have exhibited being crucial for improving the smartness and reliability of transportation systems. In recent years, a few secondary accidents occurred on the highway have attracted highly attention of traffic policemen and the supervisory aspects. To mitigate the problem, the traffic administrant aspects appeal to the R&D companies in developing a kind of portable apparatus that is used at the scene of accident and the apparatus can detect the coming cars and forecast the possible danger in advance so that the on-duty policemen have time to avoid the danger. Microwave radars have been considered to be a rational choice as the front sensing part of such collision avoidance systems and it must offer the real-time information about the distance and moving velocity of coming cars in far distance and produce the warning signal in time. This research aims at developing the DSP algorithm of radar sensing system used in such a special application. The main work is as follows.The research status of radar algorithm of detection remote target at home and abroad is investigated, and the algorithm framework in real-time and operation platform is analyzed. Meanwhile, a series of classic target capture and tracking methods are introduced and compared in this thesis, such as FFT algorithm, linear fitting algorithm, smoothing method, etc.Based on the analysis of typical target capture algorithms and tracking algorithms, radar algorithm for second collision avoidance at the scene of highway accident is proposed in this thesis. Firstly, the algorithm which is embedded in DSP (digital signal processor) operates by ping-pong data multiplexing. Secondly, the algorithm computes frequent components and phase information by radix-4 FFT algorithm. Then, the algorithm is filtrated in frequent domain and multiple targets are distinguished. Finally, the algorithm detects multiple targets with combined with tracking window and the linear fitting algorithm.According to the software platform and road experiment, this thesis will use the results of the algorithm and indicators of System requirements for algorithms, to test and verify efficiency of the new algorithm and finally demand of the application.Practical experiments show that the developed algorithm in this research is effective for avoidance of secondary collision on the highway. It makes the radar can detect and track multiple coming car objects in far distance and excite the warning signal in time. This is significant for saving life of on-duty policemen working on highway. In the same time, the innovative points in the algorithm would be usable in further software development.