Robot Navigation Technology Research Based On Visual Information

The robot navigation is the core of mobile robot, meanwhile, in the non-contact condition, vision sensors can obtain wealth of environment information, and this character of vision sensors makes the study that robot navigation based on visual information very popular.The main content of this paper is about the technology which is used to studying robot navigation which based on visual information,and MRDS that developed by Microsoft is used to modeling and simulating. Virtual mobile robot can "roaming" in the environment provided by MRDS by programming, in the process of "roaming", virtual mobile robot obtains the current environment information by some built-in sensors,for example infrared sensor,camera and compass and so on. After the end of robot "roaming",we can get the overall information of the unknown environment. Next, comparing the map that acquired by robot with the original map of the environment by Histogram.If certain conditions are met,and that environment information acquisition completed.In the process of extracting the image information of the environment,the paper used Robinson operator to extract image contour.Then,the paper stored image information and corresponding robot pose in the database in SQL Server2005.On the other hand.the positioning of the robot in an unknown environment is another problem for antonomous navigation of mobile robots.The paper used SIFT algorithm combined with Markov algorithm to solve the problem.Using Markov algorithm to predict the approximate range of mobile robot under a posture state,by that we can greatly reducing the SIFT algorithm for the calculation of the amount and did not reduce the computational accuracy. At last,the paper ultimately successful positioning of the mobile robot.Finally,in the case of mobile robot position and orientation information,the paper rasterized environment map to vertify the ability of the A*algorithm and RRT algorithm path planning.