车辆典型危险行驶状态识别与检测研究进展

发布时间：2019-04-25 17:22

【摘要】：为识别和检测车辆在行驶过程中可能出现的危险状态,及时给予驾驶员反馈和预警,使车辆始终保持安全的运行状态,重点从车辆典型危险行驶状态的识别、检测2个方面,梳理纵向及横向危险行驶状态及其表征参数,总结主要的识别与检测方法,并展望其未来研究趋势。结果表明:不同文献对驾驶事件的危险阈值划分原则差异较大,尚未形成统一的标准;隐马尔科夫模型(HMM)对危险驾驶事件的识别准确率相对较高;车辆典型危险行驶状态的几种识别方法各有优点和缺点,基于便携式设备和多传感器数据的识别方法相对较优,且基于多传感器数据融合的车辆危险行驶状态识别与检测是未来的重点研究方向。
[Abstract]:In order to identify and detect the dangerous state of the vehicle in the course of driving, give the driver feedback and early warning in time, make the vehicle keep the safe running state all the time, the recognition of the typical dangerous driving state of the vehicle is emphasized, and the detection is carried out in two aspects. Combing the longitudinal and transverse dangerous driving conditions and their characterization parameters, summarizing the main identification and detection methods, and looking forward to its future research trends. The results show that the classification principle of hazard threshold for driving events in different literatures is quite different and there is no unified standard, and the recognition accuracy of hidden Markov model (HMM) for dangerous driving events is higher than that of Hidden Markov Model (hmm). The identification methods of typical dangerous driving conditions of vehicles have their own advantages and disadvantages. The recognition methods based on portable equipment and multi-sensor data are relatively better. And multi-sensor data fusion-based vehicle dangerous driving status recognition and detection is the key research direction in the future.
【作者单位】：长安大学汽车学院;长安大学汽车运输安全保障技术交通行业重点实验室;
【基金】：国家自然科学基金资助(61374196) 教育部长江学者和创新团队支持计划项目(IRT1286)
【分类号】：U491.254

【相似文献】