Font Size: a A A

Research On Multi-robot Positioning System Based On Binocular Camera And K-means Algorithm

Posted on:2019-06-24Degree:MasterType:Thesis
Country:ChinaCandidate:Y D WangFull Text:PDF
GTID:2428330548978977Subject:Electrical engineering
Abstract/Summary:PDF Full Text Request
Robots have been used in a wide range of applications such as today's production and living areas.They are able to assist or replace people to do some work.In dangerous situations,using robots instead of humans can also protect people from injury.In this paper,a multi-robot array that deals with the leakage of hazardous chemicals is studied.The positioning system based on binocular camera and K-means classification algorithm is studied.The main research contents and results are as follows:(1)Firstly,the mobile platform adopted by HCR open source robot in this paper is introduced.It mainly includes the upper control equipment,bottom motion control board,motor and wireless network of the platform.The sensors and binocular cameras used in the positioning system are selected.According to the application scenario of the robot,a color ring-shaped identification feature with a warning color mounted on the robot is designed.(2)By analyzing the colors and geometric features of the identification features carried by the robot,a method for identifying the feature points of a single robot is proposed,and the experimental results of the recognition effect are tested both indoors and outdoors.By analyzing the conversion relationship between the camera imaging model and the image coordinate system,sensor coordinate system,camera coordinate system and world coordinate system,the problem of distortion in the camera imaging process is discussed,and a correction method for distortion is proposed.The calibrated experiment was used to solve the camera's internal parameter matrix to obtain the corrected image,and the object coordinate calculation formula was derived based on the distance measurement model.According to the epipolar constraint and parallax constraint principle,the target matching method of the binocular camera's left and right picture is studied.(3)Different classification methods were analyzed and researched.A multi-robot identification scheme was proposed.The K-means algorithm was used as the classification method of multi-robot feature points,and the computation complexity was optimized.Through the outdoor field experiment,the measurement data of the system is compared with the measured values of the ground markings,and the accuracy of the positioning system is evaluated.(4)Theoretical analysis and experimental results show that the robot identification method and the optimized K-means algorithm proposed in this paper can distinguish each robot.The robot trajectory recorded by the system can also reflect the actual movement path of the robot and achieve a positioning accuracy of 0.06 m.The results have engineering application value.
Keywords/Search Tags:multi-robot, robot positioning, stereo camera, K-means algorithm
PDF Full Text Request
Related items