未知环境下机器人自主避障行驶技术是移动机器人研究领域的核心内容。其主要目的是在外部环境未知的情况下，自主完成从起始点到目标点的无碰撞路径选择与规划，而在这个过程中可靠的环境地形数据是自主行驶的基础。而Microsoft Kinect作为一个低成本、高精度的深度传感器在近几年兴起的3D打印、虚拟现实、三维重建以及机器人导航等领域有着广泛的应用，同时也成为了众多领域研究的热门。 本文将Kinect与无人驾驶小车相结合，设计了一种小车自主避障行驶系统。该系统搭建在Windows系统上，以Kinect为深度传感器，将Kinect接收到的深度数据通过数据线发送到基于VS开发环境的软件操作平台，软件平台对Kinect接收到的深度数据做滤波处理，补充未知深度信息。通过分析处理后的深度数据的性质对障碍物进行检测并提出避障方案，进而设计出一种执行效率高，能够自主定点避障行驶的系统。该系统在操作平台上用模拟小车代替真实小车接收Kinect处理图像数据之后发出的行进指令，从而实现自主避障行驶，并带有语音导航功能，同时在行进的过程中能将障碍物信息和行走路线展示在软件平台上。在小车自主行驶过程中，该系统会将小车周围环境进行实时三维重建以便对小车周围环境有真实立体的视觉感知。可靠的深度数据是该系统成功的关键，故对Kinect深度图像数据的处理是本文研究的重点。把处理后的数据应用到自主驾驶系统中，以及路径规划设计是本文的另一研究重点。 通过测试验证小车自主避障行驶系统满足课题的基本要求，且稳定性强，为进一步扩展开发提供了借鉴作用。

Autonomous moving robots in unknown environments with obstacle avoidance study is one of the core areas of mobile robotics research.The main task is that with the external environment unknown, the robot can complete path selection from the starting point to the target point by itself , and in this process reliable depth data is the basis for independent travel. The Microsoft Kinect as a low cost and high precision depth sensor have a wide range of applications such as 3D printing, virtual reality,3Dreconstruction, robot navigation and other fields in recent years .It also become popular in many areas of research . This article combined Kinect with driverless car, we designed a car with autonomous obstacle avoidance system. The system is built on the Windows operating system with Kinect as its depth sensor,.Kinect depth data received is sent to the VS based software platform via a data line, the software will filter the Kinect depth data to add the unknown depth information . By analyzing the properties of the treated depth data to come a best idea of obstacles detecting and then design a high efficiency, capable obstacle avoidance system. The system use virtual car instead of the real car to simulate the whole autonomous driving with obstacle avoidance process .The System also has the fuction of voice navigation and the driving routes can be displayed on the software platform when the car is driving. In the car autonomous driving process, the system will display the real-time and three-dimensional reconstruction of the surrounding environment.So that can be a real three-dimensional visual perception for the people. Reliable depth data is the key to the success of the system, so the Kinect depth data processing is the focus of this study. The application of the processed data to the autonomous driving system, and the path planning is another focus of this study. The car Autonomous Driving System which is tested by experiments could meet the basic requirements of the subject, has strong stability, and provides a reference for further expansion and development.