基于视觉的智能车转向控制策略  

Intelligent vehicle steering control based on vision

作　　者：魏玉虎[1] 石琛宇[1] 姜建钊[1] 常华[1] 

机构地区：[1]中国农业大学信息与电气工程学院,北京100083

出　　处：《电子技术应用》2009年第1期130-134,共5页Application of Electronic Technique

摘　　要：在第三届全国大学生"飞思卡尔"智能车大赛参赛过程中,对提前入弯理论的提出与实践的一个总结。该策略是在获取赛道实时图像的基础上,提取引导线信息,通过对舵机理想模型的分析,创新性地利用实时检测的速度动态地改变入弯点,实现提前入弯策略,控制转向舵机,使之达到较为完美的转弯特性。实验表明,该方法提高了舵机的动态性能,实现了智能车过弯路径的最优,最终平均车速达到2m/s,入弯最大速度达到2.4m/s,弯道平均速度达到1.5m/s,满足全程高速状态的要求。This paper is a summarization of statement and practice of the enters the curved ahead of time theory in third national intelligent vehicle contest by the author. This strategy is on the base that the real time images of the lane have already been obtained , extract the information of the inlet line , make use of the speed which is real-time examinated to change the curved spot dynamicly and realizes the strategy that enter the curved ahead of time through the analysis of the servo ideal model, control the steer so as to achieve comparatively perfect turning performance and speed control. Indicated by experiments, this method improve the dynamic characteristic of the servo, realizes the turning way of the intelligent vehicle to be most superior, the average speed has reached 2m/s,the fastest curved entering speed has reached 2.4 m/s and the average curve speed has reached 1.5m/s, all of these satisfy the high speed request.

关 键 词：智能车 实时图像 舵机 引导线 入弯点 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]