Fuzzy Ranging And Map Construction Based On Image Recognition

Nowadays,smart cars and intelligent robots are developing rapidly,and detection methods based on simple vision and positioning navigation are becoming more and more important.Combining some scene obstacle detection method,we established a monocular ranging model based on fitting and fuzzy reasoning.As for a finished camera,according to the lens imaging principle,there is a continuous correspondence between the image size and the object distance.In this paper,the appropriate typical object in the space is selected to be the standard identifiable object,and the above-mentioned quantitative relationship is approximated by the fitting curve of the classic function,called the image distance function.If the closeness between the obstacle and the marker is known,the obstacle distance can be obtained by the image distance function and the imaging size of the obstacle.Moreover,higher the degree of closeness is,more accurate the ranging result is.Based on the ranging theory,an image distance function library of multiple markers under different scenes is built.In order to simplify the calculation of the ranging model matching process,we combined the specific identifier with scenes.Considering the different combinations and types of markers with in scenes,it is necessary to perform fuzzy recognition on the ranging scene to lock certain specific objects in the scene firstly.On the basis of image recognition,the recognition of the scene and its finite typical markers is determined by fuzzy inference,and then the distance of the object is calculated using the ranging method.Furthermore,we explored the error sources of fuzzy ranging,and make error analysis at the two aspects: ranging principle and numerical calculation.The existence of the threshold of the fuzzy range of the marker under the given ranging accuracy is theoretically given,and the error limit of the numerical calculation error is estimated.After the fuzzy ranging,a method for intelligent robots to map and construct simple empirical paths based on markers in an unknown environment is presented.In this paper,we utilize the linear motion as the model,making the robot obtain the direction and distance information of the same marker twice before and after,further inferring its own relative position by geometric method.After the initial positioning of the robot,we introduced a simple and intelligent experience-based search strategy.This strategy not only realizes the real-time update of the environment map,but also obtains an optimal marker map.As a result,the robot can follow a familiar process of the intelligent organism to the environment and select the optimal path within the existing experience between the markers.Furthermore,we update the map and the path in dynamic way to ensure the intelligent robot can process and avoid the obstacles automatically.Finally,low-precision camera and MATLAB development environment are utilized to carry out experiments.The results show that the fuzzy ranging method has certain accuracy,which can meet the real-time and accuracy requirements of intelligent robot ranging,and the absolute error of ranging is within 5%.Ranging can be performed in a laboratory environment,and environmental map information can be calculated and path planning can be performed.All the results show that the ranging and map intelligent construction methods are effective.