【作者】 施文武；

【导师】 陈无畏； 范迪彬； 王启瑞；

【作者基本信息】 合肥工业大学 ， 车辆工程， 2002， 硕士

【摘要】 近年来，国内外对自动引导车的研究越来越广泛。但是，比较成熟的研究是在自动引导车的路径跟踪方面；在导航方面，虽然已有很多文献提出各种方法，但是还存在很多应用条件上的限制，如：在复杂动态环境中如何避碰、如何提高自动引导车的灵活性与安全性、如何运用信息融合技术来提高自动引导车的跟踪精度等。鉴于以上原因，本文以前轮转向后轮驱动的四轮自动引导车为研究对象，建立了自动引导车的运动学和动力学模型，提出了基于安全度评判的智能导航方法和基于模糊神经网络的智能控制方法，并将该导航与控制方法分别应用运动学和动力学模型进行了仿真研究。在跟踪方面，本文提出了将被跟踪路径分成若干微小线段，应用以微小直线段代替曲线的方法，然后分段跟踪，并给出了这种方法的稳定性证明，仿真结果证明了其有很好的跟踪效果。最后，本文还应用数据融合方法对自动引导车的状态进行估计，比较了对自动引导车的状态估计使用数据融合与否对路径跟踪结果的影响。仿真结果显示了应用融合技术后自动引导车的跟踪误差可限制在较小的范围之内，与不使用融合技术的自动引导车相比，其跟踪效果更好。为验证仿真研究结果的正确性，本文还对所研究的内容进行了实验，实验结果与仿真结果基本一致，从而进一步证明了本文所提出的导航、控制、数据融合等方法的正确性。更多还原

【Abstract】 In recent years,many people have researched on AGV (Automatic Guided Vehicle) in many aspects. But the emphasis of their research is trajectory tracking or path following. On navigation,though some papers have proposed a few methods,these methods have many restrictions in application,such as how to avoid collision in a dynamic environment,how to improve the safety and agility,how to improve the precision of navigation using data fusion technology,etc. For the sake of aforementioned reasons,in this paper I take the car-like AGV as my research object,and propose a new navigation method based on safety-measure and a new intelligent control method based on FNN(Fuzzy Neural Net). Furthermore these navigation and control methods are applied to kinematic model and dynamic model in simulation to verify these methods have many fine performances. And I also propose a new method oftracking control. This method divides desired trajectory into many pieces and follows the pieces one by one. With the method,the calculation of curvature of tracking trajectory and complex calculating process can be avoided,and the performance of real-time and precision can be improved. At last,the paper proposes a method of fusing the information of relative location and absolute location for an AGV,which has good performance of high measuring precision. Simulation results demonstrate that the AGV that use the data fusion method has less error of path following than the one that does not use the data fusion method. In order to verify the correctness of the ideas proposed in this paper,we made some experiments,and the results are consistent with those of simulation,so the correctness of our ideas is verified.更多还原