Research On Key Technologies Of Indoor Positioning Based On Visual Tags

With the rapid development of computer technology,artificial intelligence and other fields,intelligent mobile platform has been widely applied to medical,warehousing and logistics,industrial production and other fields,and the demand is increasing.Among them,as one of the key technologies of intelligent mobile platform,accurate autonomous positioning is the basis of its independent work.In this paper,the indoor visual positioning system is taken as the research object,focusing on the label and its matching strategy,error characteristics,and high-precision pose solving algorithm.A low-cost and high-precision indoor visual positioning system is designed and developed.Based on the comprehensive consideration of positioning requirements and computational complexity,an indoor positioning system architecture based on visual tags is designed.At the same time,in order to ensure the real-time and robustness of label recognition process,a visual label with color and shape attributes is proposed by analyzing different label features,and the corresponding detection method is designed combined with its features,so as to reduce the complexity of label detection and the storage requirements of prior information.Due to the non uniqueness of the designed tag,it is impossible to query its corresponding coordinates directly according to the tag features.In order to solve this problem,this paper establishes the sensor sensing range model and predicts the observation label.Then,the accurate correspondence between tags and coordinates is established by using angle matching method,and then a tag search matching strategy is proposed.The experimental results show that the matching strategy can accurately match and associate tags with their corresponding coordinates.In order to improve the positioning accuracy,based on the analysis of the key factors and the error transfer relationship of the system,the positioning error model is established and the positioning error characteristics are quantitatively studied.Then,according to the error propagation characteristics,a high-precision pose algorithm based on label distribution quality identification is proposed.It can identify the geometric distribution conditions of label groups,distinguish the accuracy of state estimation,dynamically determine the weighting coefficient of label groups and solve them together.At the same time,the adaptive ability of the solution is improved while the positioning accuracy is improved.Finally,in order to test the positioning system,we design and integrate it.The simulation results show that: the positioning system designed in this paper achieves the average absolute error of 2.4cm and 6.1cm in vertical and horizontal directions,and the average absolute error of heading angle is 0.05°;The experimental results show that the positioning error of the system is in the range of 1cm ～ 18 cm,and the consumption of computing resources is low,so the system can keep running in real time on the low-cost computing platform.