Research On Robot Positioning Algorithm Based On Dynamic RSSI Database

At present,the development and application of all kinds of intelligent device are changing rapidly.As the basis,the positioning technology has obtained the unprecedented attention and research.In consideration of the wide application of Wi-Fi technology,it's a very exciting research direction to implement the robot positioning relied on the huge platform of Wi-Fi technology.Because the localization relied on Wi-Fi technology does not need extra equipment,and it just needs the existing Wi-Fi transceiver device to achieve the goal of positioning,so this system can be widely used.The thesis studied the Wi-Fi location algorithm based on location fingerprint and the updating mechanism of the location fingerprint database.By comparing and analyzing a variety of matching location algorithms,the WKNN matching algorithm based on Martensitic distance behaved well,and have get the desired results.Aiming at the problem of robot location information denoising,through the comparison and analysis of various filter fitting algorithms,we have determined the denoising method using Gaussian process regression model(GPR)to realize the interpolation,and the result is ideal.Through the system simulation,the validity of the updating mechanism of the location fingerprint data based on the monocular vision is verified.In this thesis,the main concerns are preprocessing of location fingerprint data,location fingerprint database construction,updating mechanism of location fingerprint database and matching location algorithm.During the acquisition of RSSI data,Gaussian filter has been used to preprocess the data,then get the distribution of RSSI in the location area,and then interpolate through the Gaussian process regression model(GPR).Finally,the WKNN matching algorithm based on Mahalanobis distance is used to locate.This thesis collated and analyzed the current mainstream Wi-Fi localization algorithm,and combed and contrasted the basic principles and advantages and disadvantages of the technologies involved.This thesis studies the update mechanism of location fingerprint database,and gives an engineering solution to determine whether the location fingerprint database needs updating and how to update the location fingerprint database.That is the solution of updating the location fingerprint database based on monocular vision,and the effectiveness of the solution is verified by the system simulation experiment.