【作者】 史晓磊；

【导师】 王春香；

【作者基本信息】 上海交通大学 ， 机械电子工程， 2010， 硕士

【摘要】 由于城市拥挤、泊车位紧张,辅助驾驶系统发展迅速,自动泊车系统能辅助驾驶员进行泊车,减轻驾驶员操作强度,提高泊车的安全性,具有重要的研究意义。目前的泊车系统多采用超声测距仪作为车位检测的传感器,精度有限,且不能用于垂直泊车。本文以智能车为平台,设计并实现了基于激光雷达的自动泊车系统,改进了泊车方法,可以安全顺利的完成常见的路边平行泊车及垂直泊车入库任务。实验证明此泊车系统精度高,运行稳定,维护方便。本文首先根据泊车任务需求,设计了智能车底层控制系统,这是CyberC3智能车的第三代控制系统,是实现自动泊车的基础,这代控制系统的特点是模块化。硬件上以飞思卡尔S12单片机作为核心,主要对于前轮横向控制和后轮纵向控制模块进行了设计,运行稳定,抗干扰能力强。软件上使用uC/OS-II操作系统,针对多节点的智能车系统,通讯系统以CANopen作为通讯协议,接口标准化,扩展节点方便,并且可以直接使用已经集成CANopen协议的传感器,实现即插即用,整个网络易于维护,方便上层软件的开发。然后对平行泊车的上层控制进行了实现,主要通过高精度的激光雷达来进行车位检测。泊车过程为,由人驾驶车辆经过空车位,此时对激光雷达数据处理,进行车位检测,规划路径,然后交由智能车自行完成路径跟随,来完成泊车。仿真和实验证明,车位检测结果精度高,经过改进的路径安全易控,路径跟随时选用前馈反馈结合的控制,能达到良好的控制效果。最后对于垂直泊车的上层控制进行实现。同样是通过车位检测、路径生成和路径跟随来进行泊车。其中车辆的定位采用激光雷达数据与底层传感器数据融合而得,精度更高。然后对一个简单的垂直泊车案例进行分析和实现。更多还原

【Abstract】 Drive assistance system develops quickly as the city is getting more crowded and parking space becomes tighter. The research on PAS (Parking Assistance System) has important significance because it’s able to provide assistance for the driver to reduce operating intensity and improve safety coefficient. In most PAS, Ultrasonic sensor is used to detect the parking pot so far, but it has a limited accuracy and cannot be applied in vertical parking mode. In this paper, upon the platform of the intelligent vehicle (IV), an automatic parking system based on laser radar is designed and implemented, which is capable of both parallel parking and vertical parking. Experiments show that this system is accurate, stable and easy to maintain.Firstly, according to the requirements of the parking task, the lower-level control system of intelligent vehicle is designed, which is the basis of automatic parking system. It is the 3rd generation control system of CyberC3 car, featuring good modularization. S12 singlechip is the core of the hardware. The design of longitudinal control module and the lateral control module is mainly presented, which are very stable. For the software, uC/OS-II operating system is chosen, while CANopen protocol is applied in the communication system of IV. This communication system has standard interface, and it’s easy to expand and it can directly adapt in sensors integrated with CANopen, which makes the whole system flexible and easy to maintain.Then the control method of parallel parking is presented, which detects free parking space with high-accuracy laser radar. The parking process is: the driver leads the vehicle passing by the parking pot, while the intelligent vehicle makes free parking space detection, path planning and then follows the path to complete the parking task. The paper also designs a parking method by multiple parallel shifts when there’s too little space for parking. Simulation and experiments show that: the parking space detection has high accuracy; the path planned by improved method is easy and safe to achieve; the path following method combing both forward control and backward control has a good control effect.Finally the control method of vertical parking is presented. In the same way, the parking process includes free parking space detection, path planning and path following. The localization of the vehicle is achieved by the fusion of laser radar data and lower-level sensors data with Kalman filter, making the parking process safe and accurate. Then another simple case of vertical parking is solved.更多还原