Research On The Technology Of Collision Avoidance In The ITS

In recent years, with the acceleration of urbanization, transportation networks are increasingly complex. Rear-end collision accidents happen more frequently, but how to effectively detect the occurrence of rear-end collisions becomes an urgent problem of ITS. This paper attempts to overcome this problem through the signal detection method. This paper proposes a rear-end collision detection scheme to detect the occurrence of a rear-end event. Rear-end collision detection scheme is divided into two parts: the rear-end collision model and the signal detection model.Rear-end collision model is used to detect a rear-end collision between the current vehicle and the vehicle in the behind. If a collision is found, the alarm signal has been lifted in accordance with the established model of signal until the hazardous situation. The signal detection model has always been detecting whether there is an alarm signal from the front vehicle. If the alarm signal is detected, the model immediately informs the driver to slow down or turns to park brake system. So, the target problem will be converted into how to detect the alarm signal under a high performance. For the problem, we propose two signal detection models: the signal detection model based on spatial correlation and time sequence detection model. Using the spatial correlation and temporal correlation in two signal detection models respectively, the detection probabilities for the alarm signal are increased and the detection distances in the case of the same detection probability are augmented, resulting in the more reflecting time for the driver in the following car. And a metrics, distance, is proposed to evaluate the performance of signal detection models in the paper.With the spatial correlation, the signal detection model constructs signal detectors to process the signal samples. This paper researches Covariance detector(CD), Energy detector(ED) with AND rule and OR rule. The test statistic, detection threshold and detection probability formula of these detectors are deduced. The paper also makes the theoretical analysis of the performance of ED and CD and compares their performances. The experiments verify that the detection performance of CD is over 50% higher than that of ED with AND rule and OR rule.Using temporal correlation, the detection model builds time sequence detectors, the test statistics are got by resolving the sample with time sequence detector. This paper designs three kinds of time sequence detectors: Logical union detector, Ratio union detector, Arbitration union detector. The test statistic and detection threshold formula for each detector are deduced. Finally, the paper analysis and compares the detection performance of three detectors respectively through simulation, and the conclusion can be got, in the case of 3 samples, Logical union detector can achieve the best detection performance; Ratio union detector followed; Arbitration union detector worst.Combined with the metrics of detection performance, the paper makes the performance comparison among the signal detectors of two signal detection models, and gets the conclusion that the detection probability of the detection models can be increased with the aid of the spatial correlation and temporal correlation, adding the detection distance of the models; and Logical union detector can achieve the best detection performance.