Research On Intelligent Vehicle Target Recognition Technology Based On Information Fusion

With the development of technology, the intelligent vehicle is increasingly becoming a hot at home and abroad. Environmental perception has always been a more important technology in the research of intelligent vehicle. Because of the complexity of the intelligent vehicle’s driving environment, using a single sensor to get the information can’t fully meet demand for the intelligent vehicle’s driving environmental information, thus, the information fusion technology that be able to take advantage of multi-sensor’s redundant information to compensate collaborate mutually is widely used on intelligent vehicle. This article is based on the target identification technology of the intelligent vehicle’s environment, in order to improve the accuracy and timeliness of target recognition, in the paper, we first process the millimeter-wave radar information’s information first, and then propose a Kalman filter algorithm that can able to fuse the information of V2V and other sensors, switch the observation equation depending on driving conditions. The main contents are as follows:1. The article presents an in-depth analysis of the principle of millimeter-wave radar, focusing on the Delphi ESR millimeter-wave radar which belongs to our lab, including the detection coverage, frequency, data forms and other features, while a detailed analysis of millimeter-wave radar which exist some deficiencies.2. The paper introduces the secondary development of Delphi ESR millimeter-wave radar. Including the introduction of onboard PC’s software development of millimeter-wave radar, containing real-time acquisition module, data storage module, map display module, the latter data display module, target trajectory display module.3. Completing the analysis of the raw data of millimeter-wave radar, and proposing an algorithm that can effectively eliminate the hollow signal of millimeter-wave radar, the ineffective signal and the signal which exceeds the desired range, carrying out the target’s the validity of the detection by Kalman filter to ensure the effectiveness of the target detection of the millimeter-wave radar, and verifying the feasibility of the algorithm through the experiments.4. The paper analyzes the target recognition algorithm based on information fusion and shows the analysis of the principle of the information exchange between vehicle and vehicle, assuming the existence of V2V information, which fuses with information of other on-board sensors and GPS. Article that if the information of V2V is completely correct, it is better able to accomplish the task of object recognition, but in the case of GPS or V2V communication failure, the redundant information from other sensors can assist in the intelligent vehicle for target recognition. Using the method of Kalman filtering, the kinematics model, while according to the failure of V2V or GPS to establish different observation equations, always switch observation equations, to ensure the correctness of the results, and verifying the feasibility of the algorithm through the experiments.