Design And Implementation Of Automobile Anti-collision Radar System

Automobile anti-collision radar system,as the technological base of the driver assistant system and automatic driving technique,which has a good prospect.It owns a strong weather adaptability and stable working ability,moving target indication and velocity measurement are strong points of automobile anti-collision radar,which decreases traffic accident dramatically.Automobile anti-collision radar will doubtlessly become the hot spot of marketplace in the future.Countless money can be made from it.This paper did a research on the design and implementation of automobile anti-collision radar system,and implement the system on digital hardware which based on extensible processing platform,this core is embedded with Linux,lots of development time can be saved with the operating system,and the main contents include the following:1,Introduced the basic principles of 24 GHz automobile anti-collision radar system,and detailedly derives the process of target measurement in saw-tooth wave,calculates the system working parameters according specification requirements.2,Research and design the algorithm of automobile anti-collision radar system,it is a complete signal process which from obtaining zero-IF signal to target information output.When the digital hardware gets zero-IF signal from the front-end,the first thing to do is moving target indication,and a new multiple-pulse cancellation structure is designed and verified.Next process is velocity compensation,it is hoped to move the frequency spectrum back to zero,which is shifted by the vehicle motion,the research on windowing the signal before FFT was done,several window function were compared and the suitable one was choose.In the step of signal detection,CA-CFAR and OSCFAR are two classical CFAR algorithm,they were compared and better one for automobile radar was choose.Angel scanning on uniform linear array was researched.Deep learning was tried in pedestrian-vehicle classification.Basic principles of neural network was researched.Finally,target trajectories management was designed,which based on Kalman filter to smoothness and match the trajectories,and collision-warn strategy generating.In the end,the target information and alert were output.3,Next process is program the algorithm of automobile anti-collision radar system with c-language on ARM-Linux,which based on MATLAB simulation,the main work is implementing every sub function and verify each one with data.4,Finally,test the prototype of automobile anti-collision radar system in open air,mainly,verify whether the system can go right way.On this basis,test the radar's working effects on detecting and recognizing different targets in different situation,the experiments results show that the prototype system basically meets the requirements of the project indicators.