Research On Modeling And Positioning Of Spatial Relationship Uncertainty In Indoor Locality Description

In recent years,with the development of cities and mobile Internet,people spend more and more time indoors.The demand for location based services(LBS)is increasingly diversified,and indoor location positioning has been developing rapidly.The current indoor positioning technology,which uses a variety of physical sensors to achieve positioning is complex and requires expensive equipment.It is also easily affected by the shadowing of buildings and multipath effects in a complex indoor environment.The development of cognitive technology,natural language understanding,and voice recognition technology has made the communication between people and machines more mature.Natural language,as the most basic communication method in people's daily life,has become an interactive way to gradually eliminate obstacles between people and machines.The interaction between people and machines will be more intelligent and easier for people to use.Natural language interactive via voice or text will be the future direction of location based services(LBS).Locality description exists in our daily communication.In general,locality description includes reference objects,spatial relationships,and target objects.The direction relationships and distance(quantitative and qualitative)relationships in spatial relationships can provide more locality information.In recent years,although the positioning mothed based on locality description has made great progress,there are still deficiencies in the study of spatial relation uncertainty modeling based on locality description and computation methods of complex locality description positioning scenes.In this paper,aiming at the bottleneck problem of locality description positioning,an indoor shopping market is taken as an example to study the uncertainty modeling and positioning method based on the spatial relationships in indoor locality description.The main research work and innovation results of this paper are as follows:(1)The quantitative distance model and relative direction model based on fuzzy set are proposed.Based on the analysis of people's cognitive law of quantitative distance,the quantitative distance membership function is established through statistical analysis.People's perception of spatial direction is similar to angular information.Based on this,we proposed the relative direction membership function.(2)The qualitative distance("near")model based on fuzzy set is proposed.The qualitative distance("near")is inversely proportional to the distance.The Voronoi diagram has a unique advantage in expressing the distance between points in plane.Therefore,the qualitative distance("near")membership function,which consists of the stolen-area of the Euclidean distance,is proposedbased on the fuzzy set.(3)The method based on cognitive directions and quantitative distances is proposed for positioning localities indoors with locality description.The intersetion of quantitative distance is used to determine the possible area of the target locality(i.e.,admissible domain).A joint probability function(i.e.,distance and relative direction membership functions)is presented to describe the locality probability distribution in the admissible domain.During the calculation of distance and direction,the concept of "visible segment" meet some restrictions is introduced from the viewpoint of cognition and computational efficiency.The study demonstrates that the positioning accuracy exceeds about 3.5 m within a 45 m visual space indoors.(4)The method based on cognitive directions in indoor landmark reference system(ILRS)is proposed for positioning locality with locality description.In ILRS,the nearest landmarks are possible to be selected when describing localities using direction relations indoors.Therefore,to positioning locality with direction in ILRS,the near relation should be considered.Firstly,the ILRS is proposed based on indoor shops and the qualitative distance(i.e.near)membership function is extended to make it suitable for indoor landmarks(i.e.polygon features).The possible area of the target locality(i.e.admissible domain)is determined by the neighboring area of ROs.A joint probability function that consists of qualitative distance and relative direction membership functions is provided to describe the locality probability distribution in the admissible domain.For consistency with cognition,the concept of "visible segment" meet some restrictions is provided during the calculation of relative direction,which can also reduce the number of points to be explored from an algorithmic point of view.Some cognitive experiments are conducted and demonstrate that a positioning accuracy of 3.55 m can be achieved within a 45 m visual space in ILRS.