Research And Implementation Of Self-propelled Guide Terminal Based On Indoor Positioning And Machine Vision

With the development of science and technology,wireless positioning technology is gradually entering people's lives.In the current situation that indoor positioning technology is not yet popular,indoor assistant wizard equipment with automatic follow-up and environmental identification function has become a requirement in people's life.This paper studies how to apply UWB technology with higher positioning accuracy to automatic tracking scene,how to achieve stable tracking process through data optimization processing and hardware cooperation,and how to improve the efficiency of image edge recognition in order to detect the positive direction of the environment more effectively.Finally,the wizard terminal with the function of automatic follow-up and positive direction recognition of environment is designed and implemented at the software and hardware level.The first part is the research and implementation process of the automatic following system based on ultra-wideband positioning,including hardware architecture and software control logic.The influences of various internal factors such as positioning data jitter,data arrival delay and hardware drive system limitations on the smoothness and stability of the terminal following user process are analyzed.The optimization algorithm combining sliding weighting and Kalman filtering is used.The data is processed to a smooth curve of the relative position of the user.The adaptive hardware adjustment logic is designed to automatically adjust the walking speed and the steering angle according to the moving speed and the steering range of the following user,so that the terminal can stably start,turn and follow regardless of the user's motion state.The second part is the research of the environmental direction reference line algorithm based on the improved HSI image edge recognition method.In order to enable the mobile terminal to move along the positive direction line of the indoor environment in the non-following state,it is necessary to identify the positive direction reference line of the current environment in the video stream returned by the camera.First,the color RGB image is mapped to the HSI color space,the dual-structural multi-scale morphological algorithm is used for noise filtering,and the three components of H,S and I are used for edge detection respectively,and the three-part results are weighted averaged.Due to the independencethree components,edge detection can better complete edge detection.Hough transform is performed on the edge detection result,and Gaussian weighted fitting is used to finally synthesize an environmental direction reference line.The improved edge detection algorithm can reduce the effect of sawtooth noise and ambient light and shadow on edge recognition,and more effectively identify the positive direction of the environment.The wizard terminal implemented in this paper can complete the smooth and stable follow-up process in the presence of jitter in ranging data.The improved edge recognition algorithm can obtain the edge image information more efficiently,and the design algorithm can identify the positive direction of the environment more accurately.