【作者】 赵玲；

【导师】 刘清；

【作者基本信息】 武汉理工大学 ， 控制理论与控制工程， 2007， 硕士

【摘要】 移动机器人导航技术是综合的、复杂的、不断发展的，是实现机器人智能性的核心技术。它主要包括：环境信息获取功能，环境信息理解以及实现自动导航的路径规划算法。因此移动机器人导航系统研究就是针对所处环境选择相应的传感器，然后对传感器所采集的信息特征进行处理分析，最后建立相应的路径规划控制算法。近年来，自主式移动机器人的视觉导航逐渐成为关注的热点，图像信息具有信息丰富、对场景描述全面的特点，是移动机器人感知环境的重要来源。因此本文的研究工作主要是针对移动机器人在已知环境中的路标识别算法和多传感器信息融合导航理论这两个方面的研究展开的。最后在机器人实验平台上设计并完成了移动机器人导航软件。本文首先对自主移动机器人的研究状况、发展趋势、移动机器人视觉系统在国内外的研究现状做了简要的回顾。然后介绍各种传感器的分类，如何选择合适的传感器来感知环境信息。在比较各种传感器优缺点的基础上，选择超声波传感器和CMOS视觉传感器组成机器人的传感器系统。论文介绍了CMOS传感器的工作原理，数据处理方法，并对所获得的路标图像的处理与识别进行了深入探讨，即在Hu不变矩基础上分析了离散状态下比例因子对不变矩的影响，提出了一种新改进的不变矩来提取对象的特征，从而将路标识别提升到所有交通标志识别的高度。同时本文也介绍了超声波传感器的工作原理、使用方法和数据处理方法，并用超声波传感器对移动机器人前方、左侧和右侧的障碍物进行探测；然后利用多传感器信息融合技术中的神经网络方法将视觉与超声波传感器的信息进行融合，仿真结果表明此方法可以实现移动机器人在障碍物环境下的智能导航。最后，在WiRobotX80机器人平台软件控件的辅助开发下，本文在VC++环境中设计并完成了移动机器人导航软件，实现环境地图和感知模型离线生成、电子地图、图像数据处理、路标图像识别算法，完成了在结构化环境中的导航实验。通过对实验结果和数据的进一步分析讨论，论证了所提出方法的实用性、精确性和鲁棒性。更多还原

【Abstract】 The technologies of mobile robot navigation which is comprehensive, complicated and developed continuously is the kernel technique for robotic intelligence. These technologies mainly include environment information collection, environment information understanding, environment information modeling and thereby the path planning for autonomous navigation. Therefore, the research on the navigation system of mobile robots is organized in these steps: choose the corresponding sensors for the special environment, choose the proper environment model according to the information features of sensors, and lastly build a path planning algorithm for the selected environment model. It has become a hot field to research and design the computer vision based autonomous mobile robot navigation in the past few years. Images describe roundly the environment, and carry a tremendous amount of information about the scene the robot represents. It is the important artifice for robot to explore the surroundings. The dissertation is mainly focus on robotic landmark recognition algorithm in a known environment and multi-sensor data fusion. Both the simulation result and the experiment data gained from mobile robot navigation system are provided in the dissertation.We firstly give a brief overview to the development of autonomous mobile robot and the status of motive robot vision system investigation. Then introduce the classification of various sensors, and how to choose the appropriate sensor for the environment information. Comparing each kind of sensor, we choose the ultrasonic sensor and the CMOS vision sensor to compose the robot sensor system.This dissertation introduces the principle of CMOS sensor、the method of data processing, and discusses the processing and recognition of the obtained landmark picture deeply. Namely the influences of scale factor to Hu’s moment invariants in discrete situation are investigated and present an improved algorithm of moment invariants to extract the object’s character for object recognition. Thus upgrades the landmark recognition to the level of all traffic signs recognition. Simultaneously, this dissertation also introduces the principle of ultrasonic sensor、the application method、the method of data processing, and uses the ultrasonic sensors to detect the obstacles which are in front of the robot、on the left side or on the right side, then fuses the information of vision and ultrasonic sensor with the method of Neural Network, and realizes the robot intelligent navigation through the technology of multi-sensors information fusion in the obstacle environment.Finally, mobile robot navigation software in VC++ environment is designed and realized with the help of WiRobotX80 ActiveX controller. Such essential technology as off-line constructing sensing model, map data designing, vision data process, and landmark recognizing algorithm and so on can be realized by this software. And navigation experiment in structural environment is implemented. Experiment results and further experiment data analysis show the method’s validity, robustness and practicability.更多还原