|
Target identification and location is one of the basic tasks of mobile robots, but also an important symbol of mobile robot perception and intelligent level. Vision sensor can provide a wealth of information, has become a commonly used configuration of the mobile robot, research based on visual target identification and location of the mobile robot has a very important significance. Monocular vision sensor relative to the panoramic vision sensor, vision sensor, has the advantages of simple structure, flexible movement, easy calibration. In this thesis, a monocular vision-based mobile robot target recognition and positioning algorithm, the main work done as follows: First of all, the article describes Based on the research background and significance of monocular vision mobile robot target recognition and positioning, vision-based target identification and targeting of the status quo, and monocular vision-based mobile robot target recognition and positioning algorithm framework. Second, describes commonly used several visual image feature point extraction algorithm, based on the real-time requirements of the mobile robot target recognition, target feature SURF (Speeded Up Robust Features) algorithm to extract the visual image, the use of BBF (Best Bin First) algorithm on the target feature matching, and uses the the RANSAC (RANdom SAmple Consensus) algorithm to eliminate the mismatching points, the programming vision-based target recognition, experimental results verify the algorithm to identify good effect. Third, the purpose and common methods of camera calibration. High precision for target positioning tasks require calibration, the calibration process is simple, and mature methods Zhang Zhengyou camera calibration method to achieve camera calibration. Fourth, on the basis of the completion of the target recognition, according to the target position in the current image, control the next step in the movement of the mobile robot, after exercise, to get it again contains the image of a target, for such two consecutive acquired images shot at different locations matching the target feature point, using commonly used in the field of three-dimensional reconstruction of the movement to restore the structure (Structure From Motion, SFM) algorithm, and combine the encoder information of the mobile robot, to obtain the objectives and the relative position of the mobile robot, and then from the mobile robot in the current environment in the position to achieve the goal of positioning in the current environment. Finally, a summary of the work done in this thesis, and points out the future research directions.
|