首页> 外文学位 >Implementation of a sensor fusion-based object-detection component for an autonomous outdoor vehicle.
【24h】

Implementation of a sensor fusion-based object-detection component for an autonomous outdoor vehicle.

机译:用于自动户外车辆的基于传感器融合的对象检测组件的实现。

获取原文
获取原文并翻译 | 示例

摘要

Nothing in life is certain. However the more information you have, and the better prepared you are, the more favorable the outcome. To this extent, for an autonomous vehicle to operate in the real world, it must have good information about its surroundings. For a vehicle to avoid obstacles, it must first be able to detect them. Generated data are only as good as the sensors that penetrate the vehicle's environment.; This doctoral research is the extension into the real world of the work started in Takao Okui's doctoral dissertation, Development of a Multi-Layered Map Management System Utilizing the Nonhomogeneous Markov Chain Approach. In his research, the area around the vehicle is tessellated into a local grid (the local grid map), and the range sensors are characterized by two methods. The first uses the physical properties to determine which grid cells should be updated. The second uses fuzzy modeling of the sensor's uncertainty to actually update the grid cell using a nonhomogeneous Markov chain.; Undertaking what Okui started in simulation, the techniques are implemented on a single board computer that runs on a Kawasaki Mule 500. Unlike a simulation, the constraints of the real world cannot be ignored. These include limits of storage, computational efficiency, maintainability, and the ease of adding additional, and diverse sensors.; Finally, this research provides the first implementation of the Object Detection Component (ODC) in the Department of Defense Joint Architecture for Unmanned Ground Vehicles (JAUGS). The goal of JAUGS is to present a unified architecture for the military's unmanned ground vehicles. The architecture is currently a work-in-progress, and this research will help to develop the interface and the information passed between the obstacle detection component and the requesting subsystem.*; *This dissertation includes a CD that is multimedia (contains text and other applications not available in printed format). The CD requires the following applications: Adobe Acrobat; Microsoft Office.
机译:生活中没有什么是确定的。但是,您拥有的信息越多,准备得越充分,结果就越有利。在此程度上,对于要在现实世界中运行的自动驾驶汽车,它必须具有有关其周围环境的良好信息。为了使车辆避开障碍物,它必须首先能够检测到它们。生成的数据仅与穿透车辆环境的传感器一样好。这项博士研究是对Ookui Takao博士论文利用非均匀马尔可夫链方法开发多层地图管理系统开始的工作的扩展。在他的研究中,车辆周围的区域被细分为局部网格(局部网格图),并且距离传感器通过两种方法进行表征。第一种使用物理属性来确定应更新哪些网格单元。第二种方法是使用传感器不确定性的模糊建模,使用非均质的马尔可夫链来实际更新网格单元。承接Okui在模拟中开始的工作,这些技术在运行于Kawasaki Mule 500的单板计算机上实现。与模拟不同,不能忽略现实世界的约束。其中包括存储的限制,计算效率,可维护性以及添加其他各种传感器的难易程度。最后,这项研究为国防部无人地面车辆联合体系结构(JAUGS)中的对象检测组件(ODC)提供了第一个实现方案。 JAUGS的目标是为军用无人机提供统一的架构。该架构目前仍在开发中,这项研究将有助于开发接口和障碍物检测组件与发出请求的子系统之间传递的信息。 *本文包括一张CD多媒体光盘(包含文本和其他应用程序无法使用的印刷格式)。该CD需要以下应用程序:Adobe Acrobat;微软办公软件。

著录项

  • 作者

    Novick, David Keith.;

  • 作者单位

    University of Florida.;

  • 授予单位 University of Florida.;
  • 学科 Engineering Mechanical.
  • 学位 Ph.D.
  • 年度 2002
  • 页码 p.388
  • 总页数 177
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 机械、仪表工业;
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号