【作者】 张荣辉；

【导师】 储江伟；

【作者基本信息】 吉林大学 ， 载运工具运用工程， 2006， 硕士

【摘要】 近年来,在研究高速公路智能化的同时,人们也将研究方向转向市内新型区域智能交通系统-CyberCar System的开发上来,这是智能车辆领域又一新的发展方向,是人类科技发展的一个象征。吉林大学智能车辆课题组在国内率先展开了CyberCar System的研究工作,本文在课题组自主研制的视觉引导区域交通智能车辆(CyberCar)基础上设计导航控制器,并进行试验验证。CyberCar自动转向系统复杂,采用数学解析方法通常不能反映系统特性,故采用系统辨识的方法进行建模。结合预瞄运动学模型与车辆二自由度动力学模型,构建CyberCar转向控制的状态空间模型,进而在此基础上设计线性二次型最优控制器,并针对状态变量的不可测性设计Kalman状态观测器。仿真和实验表明,CyberCar在直线导航跟踪过程中,偏差较小,最优控制器跟踪效果良好;但是在弧线导航跟踪过程中偏差很大,CyberCar转向控制的线性数学模型已经不能反映车辆控制的实际特性,最优控制器此时表现为纠偏较慢,不能满足导航控制要求。本文采用二维模糊控制结构,设计模糊控制器快速跟踪弧线导航路径,仿真和实验表明,在弧线跟踪的大偏差状态下,模糊控制器能实现快速纠偏和对导航路径的稳定跟踪。通过上述研究工作,最后结合车辆转向系统当前状态和最快动态响应能力最终建立区域交通智能车辆(CyberCar)的变结构控制输出集。试验表明,基于本文所述控制技术设计的导航控制器性能优良,能够满足CyberCar户外自主导航控制的要求。更多还原

【Abstract】 CyberCar is a subminiature and low speed automatic electromotion-vehicle.According to user’s instruction and real-time environmental condition information,it can carry on correct path planning and automatically make conduct controldecision. Then it is capable of automatically and safely driving to the destinationalong with schedule paths. CyberCar has many particular characteristics, such as nonoise, no air pollution, nimble mobile, safe and convenience, and it is the bestembodiment of “taking human as the root” in the traffic behaviors.The key technologies of CyberCar research contain many fields, mainlyincluding navigation and control technology, senor information amalgamationtechnology, communication technology etc. In recent years, vision navigation andadvanced control technology become one of the hotspot in self-navigation studiesbecause of computer image treatment technology and advanced control theory’fastdevelopment. JiLin university IV group has developed JLUIV5-CyberCar whichadopts computer vision to identify two dimensional strip paved on the ground andcarry on automatically path track. The work process of JLUIV5-CyberCar isfllowing: first of all, ground image of path information is taken by CCD;thecomputer determine vehicle’s motion direction and working state secondly;finally,tracking path steadily. With the continual improvement of people’s request to lifequality, CyberCar will have favorable application prospect in the urban traffic.So far as the outside self-navigation of the vision navigation CyberCar isconcerned, the excellent performance of the navigation controllers guarantees thestability, rapidity and reliability of the path track, which is the main point of thispaper. This paper mainly consists of five parts:1. The establishment of steering system identification and steering controlstate spatial model of CyberCar.2. The design of self-navigation optimal controller of CyberCar.3. The design of self-navigation fuzzy controller of CyberCar.4. Fuzing two controllers and build up variable structure output set.5. Simulation analysis and experimental confirmation of navigation controller.The first part of the paper briefly introduces the structure form ofJLUIV5-CyberCar, including performance parameter and main composing systemdesign. Choose adverse M sequence as system identification’input signals throughanalyzing the static state characteristic of vehicle’s steering system. The choiceprinciple, including the biggest operating frequency, clock period, sequence signalperiod, and signal peak-to-peak value, is expounded. It is show thatadverse M sequence is better than M sequence by theoretical analysis andexperiment test. Then the vehicles steering system mathematical model is obtainedby using least squares method and system identification. The simulationexperiment contrast result indicates that the mathematical model can reflect thesteering system performance well, which makes the better foundation for controllerdesign of CyberCar. Finally, the dynamic equation for the vehicle steering systemby system identification experiment is obtained, then combined with the previewkinematics model and two-degree steering dynamic model of vehicle, therefore thesteering control mathematics model based on preview kinematics was established.In the part of optimal controller, according to optimal control theory and thesteering control mathematics, we have designed the optimal controller of CyberCar.The steering controller’s optimizing parameter is the guarantee of thecontroller’s quality. Introduce the method of analysis parameters and optimizingparameter of the controller, including weight matrix Q , R and preview distance L.Design the Kalman observer to solve the measurement of the unmeasurable statevariable, φ& and v y, for JLUIV-V (CyberCar). Simulation and experimentalresults show that the controller have good robustness to the uncertain factors of thevehicle, It was proved that the optimal controller could trace the line path steadilyand reliably by the result of simulation analyses and did outdoor experiments.In order to resolve shortage of slow adjusting to big deviation of optimalcontroller during the CyberCar curve tracking, then we have design the fuzzycontroller in the following. The paper make lateral difference and orientationdifference together as one input and make its variance ratio as another input indesigning fuzzy controller. Accurate value is classified and quantified reasonablyand gauss function is chosen as degree of membership function. Fuzzycontroller’s rules are correctly made and CyberCar fuzzy controller is designed.The simulation analyses and experimental results show that fuzzy controller cantrace the curve steadily and reliably and adjust to big deviation quickly.Based on two kinds of controller feature, the paper choosesthreshold-switchmethod to carry out the fusion of controllers. By analysis onsimulation and experiment, the deviation and threshold are decided reasonable.combined with the state and the most quickly dynamic response capability of thevehicle steering system, the control output collections of intelligent vehicle tracedthe path were established. Then the variable structure control method of humansteering behavior have realized in the self-guidance of intelligent vehicle. It wasproved that the controller based on this technology could trace the path steadily andreliably by the result of simulation analyses and did outdoor experiments.更多还原