Research On Localization And Cooperation Of Multi-mobile Robots Under Distribute Vision

Multi-robot system is a key topic in robotics technology.That is an ideal platform to research multi-agent technology,and it has been widely used in swarm behaviors,system structure of multi-robot,coordination and cooperation of multi-robot and raleted topics in other same fields.Besides,the development of computer vision technology has promoted the development of intelligent robot so that robots have human's visual ability.For further study on Multi Mobile Robots System(MMRS)combined with computer vision technology,this essay launches a study on localization and cooperation problems of multi-mobile robots under global distributed vision.Firstly,vision segmentation and localizatioin is the basic of researching multi-mobile robots.On the basic of existing small mobile robots in laboratory,color label plate calibration scheme of multi-mobile robots is proposed.YUV color model is analised and selected as color label segmentation space.And aimed at the problem that a change of external lighting will have a great influence on the segmentation of mobile robot's color label plate,the influence that results from change of light can be reduced with the histogram equalization method for partial regions of robot color label image.Cluster analysis is conducted to obtain the best segmentation threshold of different robot color labels according to the distribution of circle color label pixels color value in image.A mobile robot trajectory tracking experiment is carried out to support the real-time and effection of the proposed vision segmentation and localization method.Then,aimed at the problem that the accuracy of the mobile robot vision localization is reduced by barrel distortion of the images captured by distributed vision,the calibration mothod of distored images based on Immune Network of Quadratic Polynomial Vaccine is studied.Higher calibration accuracy is obtained by vaccinating immune vaccine with a priori knowledge and iterative optimizing that combines error back propagation of BP neural network with global search advantage of immune theory.So that the calibration process is avoided that needs to calculate internal and external parameters of the camera and can not ensure accuracy.More accurate result can be obtained by he proposed calibration method and the process is easier.Finally,in the multi-robot system platform of global distributed vision,the problem of multi-robot cooperation hunting is studied based on artificial immune theory.That is applying artificial immune theory into the hunting motion control of the hunter robots.The species of the antibody population is specificed.The Farmer dynamics evolution model of the population is optimized by introduced the potential field method.The proposed method is proved by multi-robot hunting experiments in the multi-robot platform.This topic reaesrch is an exploration and practice of robots vision localization and collaboration and it has certain theoretical significance and application value.