ROS交流群
ROS Group
产品服务
Product Service
开源代码库
Github
官网
Official website
技术交流
Technological exchanges
激光雷达
LIDAR
ROS教程
ROS Tourials
深度学习
Deep Learning
机器视觉
Computer Vision

The visual positioning solution is finally complete



  • 2016.3.28
    The visual positioning solution is finally complete
    Visual positioning and navigation algorithms are the focus of current robot research, and they are also the hotspots that everyone is studying. In actual use scenarios, the need for visual positioning is also large. However, there has been no mature visual positioning navigation solution in China. Today we finally succeeded in realizing a solution for visual positioning.

    Below is a video of our test comparing the visual positioning algorithm with the general method of using the gyroscope inertial positioning.

    The test environment is Xiaoqiang (http://www.bwbot.org/zh-cn/content/xiaoqiang), a comprehensive development platform for ROS we developed. We use remote control to let Xiaoqiang repeat a square track with a length of about 1mX2m on the ground. Xiaoqiang also returns the coordinates calculated by the two groups according to Xiaoqiang’s gyroscope and camera. The above video is the position coordinates of Xiaoqiang displayed in real time on the computer. The red line represents the coordinates determined by the visual positioning algorithm, and the green line represents the small strong position coordinates determined from the gyroscope and motor encoder data. It can be seen that the results obtained by the first two methods are basically the same. As the number of turns around the gyroscope increases, the error of the gyroscope increases, but the coordinates calculated by the visual positioning do not change. This is also the advantage of a visual positioning algorithm.

    Many of today’s restaurant delivery robots still use electromagnetic or mechanical rails to position the robot. Such an approach not only has a large cost to be retrofitted, but also has a high maintenance cost, and has a great influence on the overall decoration of the restaurant. With the visual positioning algorithm, no rails are needed to aid positioning. The robot has an eye that can really be used. The same industrial AGV robot can also provide more precise coordinates to the robot by means of visual positioning.

    Update:
    After a long period of test development, we have developed a stable and complete visual navigation system. That is, the Galileo visual navigation system. More detailed information can be found here (http://www.bwbot.org/zh-hans/content/galileo).

    Introduction to Galileo system documentation (https://galileo-servicebot-doc.bwbot.org/)

    The Galileo Vision Navigation System is a robotic positioning and motion control system that incorporates multiple sensors and is guided by visual navigation. The positioning and navigation functions of the robot can be realized by loading the system. It is suitable for various application scenarios such as automatic inspection robot industry AGV, service robot and so on.

    Below is an action video that includes creating a map and saving the map. More details can be found on the Galileo system related pages.