论文部分内容阅读
智能车辆车道保持的关键在于兼顾可靠性和实时性。为此,在车道识别时,利用已识别车道分区预测临近待识别分区车道候选位置,并仅在候选位置完成后续识别以提高识别实时性;在识别各分区时,依次采用分区阈值二值化、线性滤波、基于车道宽度滤波的方法,以提高识别的可靠性。同时,在提取车道线时根据2个检测区内的车道识别结果拟合不同曲线模型的目标车道线,并进行自适应预瞄控制。采用Lab View PXI8196和数字信号处理器F2812对智能车辆车道保持系统进行了设计。道路试验结果表明,提出的车道识别及跟踪控制方法同时保证车道识别及跟踪的可靠性和实时性,且稳定性较好。
The key to maintaining a smart vehicle lane is to combine reliability with real-time performance. Therefore, in the lane recognition, the predicted lane adjacent to the lane to be identified is used to predict the position of the lane adjacent to the lane to be identified and the subsequent recognition is only performed at the candidate locations to improve the real-time identification. When the subareas are identified, the binning threshold binarization, Linear filter based on lane width filtering method to improve the recognition reliability. At the same time, the target lane lines of different curve models are fitted according to the lane recognition results in the two detection areas when the lane lines are extracted, and the adaptive preview control is carried out. The Smart Vehicle Lane Retention System has been designed with the Lab View PXI8196 and the digital signal processor F2812. The road test results show that the proposed lane recognition and tracking control method can ensure the reliability and real-time of lane recognition and tracking with good stability.