Research on Pedestrian Injury Protection Based on Multi-Objective Tracking Prediction

In road traffic, pedestrians, as vulnerable road users, usually have variable appearance attributes and random movement characteristics, and compared with drivers, pedestrians receive very limited protection. It is difficult to make a breakthrough in the improvement of pedestrian protection performance only through passive safety, therefore, domestic and foreign automobile companies and related research institutions have increased their investment and technical research on pedestrian safety. The common pop-up hood usually uses mechanical sensors to trigger the actuator, which is difficult to avoid the problems of high development cost and long cycle time, and the mechanical sensors are prone to produce undesirable effects such as non-triggering or false triggering. In this paper, we integrate the emergency collision avoidance function of advanced driving assistant systems (ADAS) and the pop-up hood system, train a large number of pedestrians and vehicles in the urban road environment through deep learning, and then utilize the camera and radar to identify pedestrians and predict the distance in real time, so as to develop a set of pedestrian protection system that integrates active and passive safety. The system can quickly trigger the hood to pop up at the T0 moment before a collision, realizing accurate protection for pedestrians, reducing development costs and false alarms while shortening the development cycle, and its comprehensive performance is more than 30% higher than that of the traditional passive pedestrian protection system, which greatly improves the protection of pedestrians.