CN106017473A - Indoor socializing navigation system - Google Patents

Abstract

The invention provides an indoor socializing navigation system. The system comprises a positioning sensor and a location server, wherein the positioning server is used for receiving first positioning sensor data transmitted by a terminal and second positioning sensor data transmitted by a target friend terminal, calculating the first geometrical coordinates of the terminal according to the first positioning sensor data and the second geometrical coordinates of the target terminal according to the second positioning sensor data, and storing the first geometrical coordinates and the second geometrical coordinates into a location database; the location server is used for acquiring indoor space model data, building an indoor space topological network, receiving a navigation request transmitted by the terminal, acquiring the first geometrical coordinates and the second geometrical coordinates from the positioning server according to the navigation request, and using a path searching algorithm to calculate the optimal navigation path in the indoor space topological network.

Description

A kind of indoor social navigation system

Technical field

The invention belongs to indoor location service technology field, particularly relate to a kind of indoor social navigation system.

Background technology

Along with mobile Internet and the development of location aware technology, location-based mobile social networking service is that people live and bring many convenience.Mobile social networking is extended to reality by position attribution therein, reduce on line the gap of real world under virtual world and line, improving the service effectiveness of social networks, people can register with position (check-in), share the content of multimedia (geo-tagging content) etc. with location tags each other；Meanwhile, user may be based on the space correlation expansion social relations of position.

At present, positional information in location-based mobile social networking is mainly by global positioning system (GPS, Global Position System), mobile base station, the location mode such as Wi-Fi obtain, corresponding service precision is at 10 meters to 100 meters, and location-based mobile social networking service precision is in building building rank mostly.And owing to the time of everyone every day up to 90% average is in indoor, people also have great demand for indoor location service, the indoor navigation between such as good friend.But, in prior art, the positional information in mobile social networking and the service relevant to position do not become more meticulous the functional space of indoor many floors, and location-based mobile social networking service belongs to the services of user's on-demand mostly, lack the real-time actively updated, cause can not the most accurately planning guidance path when indoor social navigation.

Based on this, need badly at present and a kind of there is high accuracy, real-time indoor social navigation system.

Summary of the invention

The problem existed for prior art and the deficiency of application, embodiments provide a kind of indoor social navigation system, for solve indoor social connections navigation of the prior art seek friend's function can not in real time, dynamic, plan to accuracy the technical problem of guidance path.

The present invention provides a kind of indoor social navigation system, and described system includes:

Location-server, for receiving the first alignment sensor data and the second alignment sensor data of target good friend's terminal transmission that terminal sends, calculate the first geometric coordinate of described terminal according to described first alignment sensor data, calculate the second geometric coordinate of described target terminal according to described second alignment sensor data；And described first geometric coordinate and described second geometric coordinate are stored in location database；

Location server, is used for obtaining indoor location model data, builds the space topological network of each level of indoor location model according to described interior space position data；

Receive the navigation requests that described terminal sends, obtain described first geometric coordinate and described second geometric coordinate according to described navigation requests to described location-server, and utilize path search algorithm to calculate optimum guidance path in described space topological network.

In such scheme, the space topological network of the described each level of indoor location model includes: fine granularity level AEGVG figure, outlet level illustraton of model and position level illustraton of model.

In such scheme, the fine granularity level AEGVG figure that described location server builds indoor location model according to described interior space position data specifically includes:

Extract one-dimensional skeleton according to indoor floor plan, form the one-dimensional Voronoi diagram in interior space elongated zones；

Described open area is carried out stress and strain model with the default length of side and forms grid chart, described grid chart is added in described Voronoi diagram；

Carry out sampling node with the average step length of pedestrian for the sampling interval, generate described AEGVG figure.

In such scheme, the outlet level illustraton of model that described location server builds indoor location model according to described interior space position data specifically includes:

The Egress node of described coarser grained layers is determined according to the exit position in described fine granularity layer AEGVG figure；

Reachable path between adjacent position is built as limit described outlet level illustraton of model.

In such scheme, the position level illustraton of model that described location server builds indoor location model according to described interior space position data specifically includes:

The nodes of locations of described coarser grained layers is determined according to the character position in described fine granularity layer AEGVG figure；

Described position level illustraton of model is generated according to adjacent, the connected relation between described nodes of locations.

In such scheme, described location-server, for calculating the first geometric coordinate of described terminal according to described first alignment sensor data, specifically includes:

When described location-server detects the anchor point signal in described first alignment sensor data, described anchor point signal is carried out fingerprint matching with location fingerprint data base and determines the initial position of described terminal；

Detect described anchor point signal with default cycle timing again, if described anchor point signal being detected, utilizing particle filter to merge location algorithm fusion pedestrian's dead reckoning PDR method, described anchor point signal and interior space information and determining the first geometric coordinate of described terminal.

In such scheme, after described location server builds the space topological network of each level of indoor location model, specifically it is additionally operable to:

Receive the factor of influence in every bar navigation path in described space topological network；

Receive each described factor of influence weighing factor to current navigation path；

The comprehensive weights of every paths are calculated according to described weighing factor.

In such scheme, described factor of influence specifically includes: indoor pedestrian's reach distance, up to time, density of personnel and road width.

In such scheme, described indoor pedestrian's reach distance is by formulaCalculate；Wherein, described O_iFor described terminal corresponding first move object；Described O_jFor described target good friend's terminal corresponding second move object；Described (x_x,y_k) it is that fine granularity layer first moves object O described in distance_iNearest node n_kCoordinate；Described m is integer.

In such scheme, the evaluation function of described path search algorithm is: f (n)=g (n)+h (n)；Wherein, described f (n) is the start node evaluation function via node n to destination node；Described g (n) is that start node described in state space is to the actual cost of node n；Described h (n) is the node n actual cost to the optimal guidance path of described destination node.

The invention provides a kind of indoor social navigation system, described system includes: location-server, the second alignment sensor data that described location-server sends for the first alignment sensor data and target good friend's terminal receiving terminal transmission, calculate the first geometric coordinate of described terminal according to described first alignment sensor data, calculate the second geometric coordinate of described target terminal according to described second alignment sensor data；And described first geometric coordinate and described second geometric coordinate are stored in location database；Location server, is used for obtaining interior space model data, builds interior space topological network；Receive the navigation requests that described terminal sends, obtain described first geometric coordinate and described second geometric coordinate according to described navigation requests to described location-server；Path search algorithm is utilized to calculate optimum guidance path in described interior space topological network；So, described location server can obtain the geometric coordinate of navigation both sides from location-server；Build the topological network of each level of interior space position model, and geometric coordinate is mapped in the fine granularity layer of described locus model, determine the factor of influence in every bar navigation path, each factor of influence weighing factor to guidance path in topological network, calculate the comprehensive weights of every paths according to weighing factor, determine optimum guidance path according to comprehensive weights；Thus can in real time, dynamically, accurately plan guidance path.

Accompanying drawing explanation

The overall structure schematic diagram of the indoor navigation system that Fig. 1 provides for the embodiment of the present invention one；

The interior space position model HiSeLoMo frame construction drawing that Fig. 2 provides for the embodiment of the present invention one；

The one-dimensional skeleton drawing of the indoor plane figure that Fig. 3 provides for the embodiment of the present invention one；

The fine granularity layer AEGVG figure of the described HiSeLoMo that Fig. 4 provides for the embodiment of the present invention one；

Position level position model schematic diagram in the coarser grained layers that Fig. 5 provides for the embodiment of the present invention one；

The coarser grained layers middle outlet hierarchical position model schematic that Fig. 6 provides for the embodiment of the present invention one；

The mobile object dynamic relationship of topology schematic diagram that Fig. 7 provides for the embodiment of the present invention one；

The interior space position model HiSeLoMo interlayer relation schematic diagram that Fig. 8 provides for the embodiment of the present invention one

Attribute Association relation schematic diagram between the interior space each layer of position model HiSeLoMo that Fig. 9 provides for the embodiment of the present invention one.

Detailed description of the invention

In order in real time, dynamically, accuracy plan under indoor environment the guidance path between good friend, the invention provides a kind of indoor social navigation system, described system includes: location-server, the second alignment sensor data that described location-server sends for the first alignment sensor data and target good friend's terminal receiving terminal transmission, calculate the first geometric coordinate of described terminal according to described first alignment sensor data, calculate the second geometric coordinate of described target terminal according to described second alignment sensor data；And described first geometric coordinate and described second geometric coordinate are stored in location database；Location server, is used for obtaining interior space model data, builds interior space topological network；Described first geometric coordinate and described second geometric coordinate is obtained to described location-server according to described navigation requests；Path search algorithm is utilized to calculate optimum guidance path in described interior space topological network.

Below by drawings and the specific embodiments, technical scheme is described in further detail.

Embodiment one

The present embodiment provides a kind of indoor social navigation system, as it is shown in figure 1, described system includes: terminal 1, target good friend's terminal 2, location-server 3, location server 4, social networking application server 5；Wherein,

Before described terminal 1 is wanted to carry out good friend's real-time navigation function, described location server 4 is for calculating the distance between described terminal 1 and each good friend's terminal；Described social networking application server 5 is additionally operable to the distance according to distance and described good friend's terminal (with tabular form) is shown on the interface of described terminal 1.

Here, the most described terminal 1 selects to send navigation requests to target good friend's terminal 2, and after described navigation requests is permitted, described navigation requests is forwarded to described location server 4 by described terminal 1.

After described navigation requests obtains the license of described target good friend's terminal 2, described location-server 3 receives, for the cycle preset with first, the first alignment sensor data and the second alignment sensor data of described target good friend's terminal 2 transmission that described terminal 1 sends, to calculate the first real-time geometric coordinate and the second real-time geometric coordinate of described terminal 1 and described target good friend's terminal 2.Wherein, the described first cycle preset was 1HZ.

Specifically, when described location-server 3 determines the first geometric coordinate of described terminal 1 according to described first alignment sensor data, it is first determined whether the anchor point signal detected in the first alignment sensor data, if be detected that during anchor point signal, received signal strength value according to described anchor point signal, use arest neighbors matching algorithm that with location fingerprint data base, described anchor point signal is carried out location fingerprint to mate, distance between signal calculated intensity level and each finger print data of location fingerprint data base, therefrom choose the finger print data that minimum range is corresponding, using its representative geometric coordinate as the initial position of described terminal 1；If be not detected by anchor point signal, the blind area point of GPS/ base station signal, the alignment sensor data of described terminal 1 are successively selected to present the characteristic point of special state and the mode of (map reconnaissance, sweep Quick Response Code etc.) that interacts with described terminal 1 to determine initial position according to priority.Wherein, the characteristic point of described special state is that the change of described alignment sensor data is more than data during predetermined threshold value.

Here, after location-server 3 determines the initial position of terminal 1, initial position data is stored to location database.Wherein, described signal strength values is to be given by the WiFi/ Bluetooth signal receiver module measurement of described terminal 1；Described first alignment sensor data may include that acceleration, angular velocity and towards；Described anchor point signal may include that Wi-Fi signal or Bluetooth signal.

After location-server 3 determines the initial position of terminal 1, utilize pedestrian dead reckoning (PDR, Pedestrain Dead Reckoning) method carries out the real-time geometric coordinate of terminal 1, simultaneously with the second cycle timing detection anchor point signal preset, utilize particle filter to merge location algorithm and merge the multi-source information such as characteristic point of the characteristic point of anchor point signal, interior space information (indoor map) and the first alignment sensor data, with the location cumulative error in correction PDR procedure further, so that it is determined that the real-time geometric coordinate of described terminal 1.Wherein, depending on the described second cycle preset can be according to the configuration of terminal 1, it is traditionally arranged to be 10～20HZ, it is preferable that for 11HZ, 12HZ, 15HZ, 18HZ or 19HZ；The characteristic point of described anchor point signal is described anchor point signal anchor point signal strength values when undergoing mutation.Described indoor map includes: interior space elements position and the structures thereof such as wall, room, corridor, door.

Specifically, described location-server 3 utilizes the multi-source informations such as particle filter fusion location algorithm PDR method, anchor point signal and interior space information to determine that the first real-time geometric coordinate of described terminal 1 specifically includes:

The state vector holding the mobile destination object to be positioned of terminal 1 is designated as X by described location-server 3_i=(x_i,y_i,α_i)^T, i=1,2 ..., N, wherein, (x_i,y_i) denotation coordination, a_iParameter for Weinberg step-length model.Then, particle filter merges shown in the sensor model such as formula (1) of location algorithm, shown in motion model such as formula (2):

$Z_k={(a_k,{\mathrm{\θ}}_k^{\′},{\stackrel{\·}{\mathrm{\θ}}}_k)}^T+\mathrm{\η}---\left(1\right)$

Wherein, in formula (1), a_kFor the acceleration of kth step, described acceleration a_kCan be drawn by the accelerometer measurement in terminal 1；θ_k' for kth step towards, described towards being drawn by the lining of terminal 1,For the angular velocity of kth step, described angular velocityCan be obtained by the gyroscope measurement in terminal 1；η represents Gaussian random process.

$x_k^i=x_k^{i-1}+s_k^{i-1}{\mathrm{sin\θ}}_k^i---\left(2\right)$

$y_k^i=y_k^{i-1}+s_k^{i-1}{\mathrm{cos\θ}}_k^i---\left(3\right)$

Wherein, in formula (2),It is to be a by parameter_iWeinberg step-length modelCalculate.For kth step i-th particle towards, can be calculated by formula (4):

${\mathrm{\θ}}_k^i=kf(H\·Z_k)---\left(4\right)$

Wherein, in formula (4),The result calculated according to compass and gyroscope measured value for Kalman filter.Particle filter merges the concrete calculation procedure of location algorithm and can be described as follows:

A) initialize: i.e. calculate the initial position of target according to signal strength values, according to the measured value of compass determine target towards.

B) prediction: obtain the state of k moment N number of particle according to the motion model of target

C) weight computing: have two kinds of situations to need to recalculate weights.The first situation, the particle weights through wall or barrier is assigned to 0；The second situation, when running into characteristic point, particle weights will recalculate according to the distance with characteristic point, and distance is nearer, and weight is the biggest.In the present invention, compose bigger weights by the particle nearer to distance feature point, can reach to revise the effect of position error, also can keep good Consumer's Experience simultaneously.The computing formula of weights is as shown in (5).

$w_k^i=w_{k-1}^i\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}\exp (-\frac{\left|\right|X_{z_k}-X_{x_k^i}|{|}^2}{2{\mathrm{\σ}}^2})---\left(5\right)$

Wherein, in formula (5),Being characterized coordinate a little, σ is corresponding standard deviation.

Additionally, when environment exists wireless signal (Wi-Fi, bluetooth), and location-server 3 is by resolving the alignment sensor data of described terminal 1, (present in finger ring border, some do not have the characteristic point of distinguishing mark to capture characteristic point, such as magnetic anomaly point and corner etc.), the positional information then obtained characteristic point and wireless fingerprint is weighted averagely, and more new particle weights；If only perceiving wireless signal, then utilize the location updating weights that wireless fingerprint obtains；When only capturing characteristic point, then the position that distinguished point based produces, update weights.Wherein, described wireless fingerprint is signal strength values mentioned above；Described positional information is real-time geometric coordinate；After calculating the weights of particle, need according to formula (6), weights to be normalized:

${w_k^i}^{\′}=w_k^i/{\mathrm{\Σ}}_{j=1}^N{w}_k^j---\left(6\right)$

Wherein, in formula (6),Represent the weight of k moment i-th particle,Represent the weight sum of all particles.

D) state estimation: probability distribution over states p of filtered mobile target_k(x_k|y_1:k) can approximate representation be:

$p\left(x_k\right|z_{1:k})\≈{\Σ}_{i=1}^N{w_k^i}^{\′}\mathrm{\δ}(x_k-x_k^i)---\left(7\right)$

And thus can draw the estimation of location status, as shown in formula (8):

${X_k}^{\′}={\mathrm{\Σ}}_{i=1}^N{w_k^i}^{\′}{x}_k^i---\left(8\right)$

E) resampling: the basic thought of resampling is to replace, with the particle that weights are big, the particle that weights are little.When causing sample number deficiency owing to eliminating invalid particle, need to carry out resampling according to the information of previous moment, be now not required to update Weinberg step-length model parameter a_i。

F) correction.Measured value according to each sensor, it is judged that whether target arrives the vicinity of characteristic point.If reaching the near zone of certain characteristic point, just according to the position of described feature point pairs target and towards being modified, circulation performs (b)-(f).

And when described location-server 3 does not detects described anchor point signal with the second cycle preset, the real-time geometric coordinate of terminal is then determined according to described PDR method, specifically: described location-server 3 is on the basis of determining the current geometric coordinate of user terminal, the walking event of capture user, the step-length of pedestrian's walking is calculated according to accelerometer, according to compass determine pedestrian towards, retrained by interior space information (indoor map), calculate next step position of user, and then determine the first real-time geometric coordinate of described terminal 1, when described location-server 3 detects described anchor point signal, the real-time position information of the terminal 1 of pedestrian's dead reckoning method estimation is corrected, to reduce the cumulative error of pedestrian's dead reckoning PDR method.

Here, when described location-server 3 determines the second geometric coordinate of described target good friend's terminal 2 according to described second alignment sensor data, identical with the method for the first geometric coordinate determining terminal 1, do not repeat them here.

Further, after described location-server 3 determines described terminal 1 and the first real-time geometric coordinate of described target good friend's terminal 2 and the second real-time geometric coordinate, described location server 4 is used for obtaining interior space position model data, builds the space topological network of each level of indoor location model according to described interior space position data.

Specifically, the space topological network of the described each level of interior space position model includes: fine granularity level AEGVG figure, outlet level illustraton of model and position level illustraton of model.The space topological network that described location server 4 builds each level of indoor location model according to described interior space position data specifically includes:

According to enclosed spatial characteristic and mobile destination object motion feature, build the fine granularity layer AEGVG figure of described interior space position model HiSeLoMo based on indoor floor plan, determine the geometric coordinate of interior space object, character position, topological relation and time-space relationship semantic information.Wherein, institute's semantic information can be particularly as follows: the attribute such as proximity relations between room and the connected relation in corridor, room, the mobile geometric coordinate of object, character position (room number), function, space-time restriction.Wherein, the framework of described indoor interior space position model HiSeLoMo is as shown in Figure 2.

Specifically, the fine granularity layer AEGVG figure of described HiSeLoMo includes: the one-dimensional Voronoi diagram in interior space elongated zones and open area two dimension rule coverage grid chart.Generally, interior space elongated zones is expressed by one-dimensional Voronoi diagram, and open area then utilizes grid chart to express.Wherein, when the width of interior space unit is referred to as elongated zones less than or equal to certain value (such as 3m) region, such as corridor etc.；It is open area, such as hall etc. when the width of interior space unit is more than the region of certain value (such as 3m).

Here, the fine granularity layer AEGVG map generalization of described HiSeLoMo specifically includes:

First, extracting one-dimensional skeleton according to described indoor floor plan, form Voronoi diagram, described one-dimensional skeleton is as shown in Figure 3；Open area is carried out stress and strain model with the default length of side and forms grid chart, described grid chart is added in described Voronoi diagram；Carrying out sampling node with the average step length of pedestrian for the sampling interval, create described AEGVG figure, described AEGVG schemes as shown in Figure 4.Wherein, carry out sampling node using the average step length of pedestrian as the length of side, meet the motion feature of pedestrian, it is possible at utmost reducing the number of nodes in model, described pedestrian's step-length is about 1m.Simultaneously, it is contemplated that the step-length of people's walking is at about 1m.Therefore, to open area, then square net with the length of side as 1m divides, and builds open area graph model based on this.

Here, the fine granularity layer AEGVG graph model of described HiSeLoMo can carry out formal definitions according to formula (9):

G_fine=(V_fine,E_fine) (9)

In formula (9), V_fine={ v_i, it is the set of described AEGVG figure interior joint；It it is the set on limit in described AEGVG figure；Each edge is made up of two nodes, by formula (10) Suo Shi.

E=(V_i,V_j) (10)

Wherein, each nodeEach node describes a certain discrete location of the interior space, has the attributes such as geometric coordinate, state, label；Generally, the attribute information of described node can pass through v_id,x_v,y_v,c_v,s_v,l_v,f_v,b_v> represent.Described v_idIt it is numbering ID of described node；Described (x_v,y_v) it is the geometric coordinate of node；Described c_vFor the space type of described node, described c_v∈{room,corridor,door,vertical,passage}；Described s_vFor the physical state of described node, described s_v∈ { free, occpuied}, described l_vFor the tag attributes of node, described f_vFloor identification residing for described node, described b_vBuilding mark residing for described node.

Further, described limit e ∈ E_fine, have expressed the connected relation of each node in AEGVG figure, the attribute on limit is < e_id,v_i,v_j,f_e,b_e,w_e>, wherein, v_i,v_jRepresent two end nodes on limit, f_eAnd b_eRepresent the character position attribute on limit, i.e. floor corresponding to limit and building information.While there may exist the membership relation of one-to-many, i.e. one limit have passed through multiple functional space unit.Described w_eRepresent the weight on limit, generally using the Euclidean distance of two nodes as weighted value.

Secondly, position hierarchical model is built；Specifically, on the basis of fine granularity layer AEGVG graph model, take out the position hierarchical model of coarseness.Here, position level, by the organizational form of a kind of level, expresses the semantic informations such as the topological relation (such as adjacent, inclusion relation) between object and time-space relationship (time-space matrix, space-time restriction etc.).Generally, position is divided into three major types: room (Room), vertical lift space (Vertical Passage), including stair, elevator etc.；Corridor (Corridor).Here level refers to adjoining up to ordering relation between position, and such as: sequentially pass through which adjoining position from certain entry position, these adjoining positions are as the child node of entry position in hierarchy chart；Or the space inclusion relation between position, such as: which position certain floor comprises, these positions are as the child node of hierarchy chart.

AEGVG graph model based on fine granularity layer, will have same label attribute l_vNode aggregation be a character position.The nodes of locations of described coarser grained layers is determined according to described character position；After nodes of locations in forming coarser grained layers, according to adjacent, the connected relation between nodes of locations, so that it may form position hierarchical model complete in coarser grained layers.Position hierarchical model is generally with node on behalf character position, while represent that position adjoins or the hierarchical graph model of inclusion relation, and can be as shown in formula (11).

G_loc=(V_loc,E_loc) (11)

In formula (11), V_lo_c={ v_i, represent the set of all character positions；Represent that in AEGVG figure, position adjoins or the set of inclusion relation；Each edge e_loc=(v_i,v_j∈E_ioc).Meanwhile, each character position v_i=< loc_id,c_loc,l_loc,f_loc,b_loc, adj_loc >, described loc_idFor the numbering of abstract position space, c_locFor the classification of abstract position space, described c_loc∈ { room, corridor, vertical passage}, l_locRepresent the sign semantic information of abstract position space；f_locRepresent floor residing for abstract position space；b_locRepresent the building of abstract position space；Meanwhile,It is, with notional position, there are all location sets of neighbouring relations.

In reality, as a example by certain engineering Lou Si building, 4th floors fine granularity floor AEGVG figures being carried out abstract forming position node, as it is shown in figure 5, room location circular node represents, vertical lift locus is represented by square nodes, and corridor is then represented by triangular nodes.Such as, the fine granularity node in the vertical space VP2 in fine granularity layer, corridor section HW4 and room RM12 is abstracted into nodes of locations VP2 in coarser grained layers, HW4 and RM12 respectively.After nodes of locations in forming coarser grained layers, according to the relation between nodes of locations, forming position level, as shown in Fig. 5 lower left.Such as nodes of locations VP2 is connected with corridor section node HW4, and HW4 is connected with corridor node HW5, and the nodes of locations such as HW5 with RM14, HW6 is connected or adjacent.By adjacent, the connected relation between nodes of locations, so that it may form position hierarchical model complete in coarser grained layers.

Then, the Egress node of described coarser grained layers is determined according to the exit position in described fine granularity layer AEGVG figure；Reachable path between adjacent position is built as limit described outlet hierarchical model.

Specifically, in conjunction with the position hierarchical model of HiSeLoMo coarser grained layers, in order to the spacing and topology supporting the position of coarseness is expressed, on the basis of fine granularity layer model, the outlet hierarchical model of coarseness is taken out.Here, export the level organizational form by a kind of level, express the semantic informations such as the topological relation (such as connected relation, ordering relation) between exit position, distance, constraint.Wherein, outlet refers to two junction points up to locational space in communication chamber, including actual outlet and virtual outlet two class.Actual outlet be two space cells up to gateway, usually room door；And virtual outlet be according to subspace unit between connected relation and artificially defined gateway, do not exist in doors structure.One outlet can only connect two locational spaces, and a space cell can comprise multiple outlet, and outlet is the unique channel connecting different spaces unit.Level then points out the connected relation (as certain exit position has connected two locus) between mouth, or passed through during pointing to reach certain exit position outlet ordering relation (as from certain floor exit arrive the outlet of certain position the ordering relation of outlet of process).

Connecting the Egress node set between different spaces unit in outlet level correspondence fine granularity layer, this set is according to the category attribute c in space in fine granularity layer AEGVG model_vNode extraction for outlet obtains.Egress node is according to the syntopy in space (arriving at order) cambium layer aggregated(particle) structure, and wherein top-most node indicates entry into the entrance in this space, from top mode down, the node of different layers represent up to hierarchical sequence relation.As shown in Figure 6, the Egress node DR57 that in certain engineering Lou Si building plane graph, VP2 region is corresponding is top mode, can arrive DR55 Yu DR20 two outlet, and therefore two Egress nodes of DR55 and DR20 are as the child node of DR57.

By position exports the abstract Egress node for coarseness, the reachable path between adjacent position, as limit, builds outlet hierarchical model.Described outlet hierarchical model can be represented by formula (12).

G_exit=(V_exit,E_exit) (12)

In formula (12), V_exit={ v_iIt is the set of all outlet ports node, described Egress node can be represented by formula (13).

v_i=< ex_id,l_ex,loc_i,loc_j,parent_ex> (13)

In formula (13), ex_idRepresent the numbering of Egress node, keep consistent with the node serial number that fine granularity sheaf space type is door, l_exRepresent the semantic locations information of Egress node, the functional attributes in space as represented by node.Exit position is generally connected to the position of two connections, and the position of two connections is by (loc_i,loc_j) represent.loc_iAnd loc_jTwo positions in correspondence position level respectively, said two position refers to any two node in the level of position.Described parent_exRepresent Egress node father node numbering in outlet hierarchical tree structure,And E_exit=V_exit×V_exitBeing the set of all reachable paths, every paths can pass through formula (14) and represent:

e_exit=v_i×v_i (14)

Wherein, e_exit∈E_exit。

Further, the mobile object layer model of described interior space position model HiSeLoMo is built.Specifically, because, in mobile computing environment, there is substantial amounts of mobile object (such as personnel, mobile asset etc.).For convenience, described mobile Object table can be shown as Moving ObjID, (x, y, t), objsemantic >；Wherein,

Described MovingObjID is the numbering of described mobile object, described, and (x, y, be t) the t geometric coordinate that moves object, and described objsemantic is the semantic information of mobile object.

Here, if Σ is objsemantic={ Σ person ∪ Σ asset},

Then objsemantic ∈ ∑ objsemantic={person_id, asset_id}.

In order to simplify the dynamic relationship of topology between mobile object, topological diagram G based on HiSeLoMo fine granularity layer_fine, mobile object MovingObject is mapped at the topological relation of certain moment t the topological diagram G of fine granularity layer_fine-sub, as shown in Figure 7.Concrete expression way is: according to described mobile object MovingObject, in the position of certain moment t, (x, y), at fine granularity etale topology figure G_fineThe node NearestNode that middle inquiry is nearest apart from this position；Described mobile object MovingObject is i.e. represented by the fine granularity etale topology subgraph G at NearestNode place at the topological relation of moment t_fine-sub.Wherein,Then MovingObject_i,tWith NaerestNode_iMapping mutually, mapping relations can be represented by formula (15).

f:MovingObject_i,t→NearestNode_i (15)

Finally, the interlayer relation of described interior space position model HiSeLoMo is determined.

Specifically, in coarser grained layers, position hierarchical model can be polymerized from fine granularity layer and obtains, and outlet hierarchical model can derive from fine granularity layer, position level and outlet layer secondary between can also mutually derive, as shown in Figure 8.Owing to an outlet is connected to two adjacent spaces, contain this connection or proximity relations, so can mutually derive between outlet layer and site layer at position level in outlet level.Relation on attributes in fine granularity layer and position level and outlet level is as it is shown in figure 9, node, the attribute on limit in outlet layer and site layer are all derived from fine granularity layer.

Further, after described location server 4 builds the space topological network of each level of indoor location model, specifically it is additionally operable to receive the factor of influence in every bar navigation path in space topological network；Receive each described factor of influence weighing factor to current navigation path；The comprehensive weights of every paths are calculated according to described weighing factor.Wherein, described factor of influence specifically includes: indoor pedestrian's reach distance, up to time, density of personnel and road width.

Specifically, optimum guidance path is mainly calculated for factor of influence by the present embodiment with indoor pedestrian's reach distance, therefore, the reach distance of indoor road is the weights in path, then indoor pedestrian's reach distance based on constructed indoor location model can be calculated by formula (16):

$IOD(o_i,o_j)=\underset{k=1}{\overset{m}{\&Sigma;}}\sqrt{{(x_{k+1}-x_k)}^2+{(y_{k+1}-y_k)}^2}---\left(16\right)$

Wherein, in formula (16), described IOD (O_i, O_j) it is indoor pedestrian's reach distance；Described O_iFor described terminal 1 correspondence first moves object；Described O_jFor described target good friend's terminal 2 correspondence second moves object；Described (x_x,y_k) it is that fine granularity layer first moves object O described in distance_iNearest node n_kCoordinate；Described m is integer.

Here, while described location server 4 builds the space topological network of the described each level of indoor location model, described terminal 1 and the first real-time geometric coordinate of described target good friend's terminal 2 and the second real-time geometric coordinate is obtained to described location-server 3, path search algorithm is utilized to calculate optimum guidance path in described space topological network, wherein, described optimum guidance path is the path that indoor pedestrian's reach distance is the shortest, and the evaluation function of described path search algorithm is: f (n)=g (n)+h (n)；Wherein, described f (n) is the start node evaluation function via node n to destination node；Described g (n) is that start node described in state space is to the actual cost of node n；Described h (n) is the node n actual cost to the optimal guidance path of described destination node.The present embodiment is using node n to described destination node between Euclidean distance as weighted value, specifically comprising the following steps that of its route searching

(1) the fine granularity layer node being mapped in described indoor location model with the second real-time geometry left side described in target good friend's terminal 2 by the described first real-time geometric coordinate of the terminal 1 of navigation both sides, start node and goal node are all respectively v_startAnd v_goal。

(2) by described start node v_startPut into (the f value of described open list and g value are all 0) in open list OPEN.

(3) at v_startLocation unitary space loc_startStart Path extension search, OPEN searches the node with minima, and using the node found as current node.

(4) current node is deleted from OPEN, current node is added and closes list CLOSE.

(5) each node that current node is adjacent is performed successively step (6) (8), as goal node v_goalWhen being added into open list as node to be tested, represent and searched path, now end loop；Or work as loc_startCorresponding outlet node v_exit-sWhen being put into open list as node to be tested, represent at current location element space loc_startDo not search path, be now switched to outlet layer from fine granularity layer and carry out Path extension search, will outlet node v_exit-sDelete from OPEN, put into closing list CLOSE, and perform step (9).

(6) if this neighborhood of nodes impassabitity or in CLOSE, then continue to extend next node.

(7) if this neighborhood of nodes is not in OPEN, then this node is added in OPEN, and the father node of this neighborhood of nodes is set to current node, preserve g value and the f value of this neighborhood of nodes simultaneously.

(8) if this neighborhood of nodes is in OPEN, if then judging whether to arrive the g value of this neighborhood of nodes less than the original g value preserved via current node, if being less than, then father's node of this neighborhood of nodes is set to current node, and resets g value and the f value of this neighborhood of nodes.

(9) with v_exit-sAt outlet layer, each of which adjacent outlets node is performed step (6) (8) for current node, as goal node v_goalLocation unitary space loc_goalCorresponding outlet node v_exit-gWhen being added into open list as node to be tested, it is switched to fine granularity layer from outlet layer and carries out Path extension search, will outlet node v_exit-gDelete from OPEN, put into closing list CLOSE, and perform step (10).

(10) with v_exit-gReturn to fine granularity layer fine granularity adjacent to each of which layer node for current node and perform step (6) (8), as goal node v_goalWhen being added into open list OPEN as node to be tested, represent and searched path, now end loop；Or when for empty, showing, without the new node that can add, the node of inspection does not has goal node v_goalThen mean that path cannot be found, the most also end loop.

So far, described optimum guidance path determines, described optimum guidance path is shown on the interface of described terminal 1 by described location server 4, simultaneously, in navigation procedure, terminal 1 and target good friend's terminal 2 can also look for the mutual and contact during people by text chat increase, with the information lacked in supplementary map navigation procedure.

The indoor navigation system that the present embodiment provides, when binding hierarchy indoor location model and path search algorithm determine optimum guidance path, it is possible to decrease the complexity of algorithm, improves search efficiency and navigation accuracy.

Embodiment two

Corresponding to embodiment one, after the real-time geometric coordinate of terminal 1 may determine that, it is possible to issue, according to described geometric coordinate, content of registering in described social networks；Specifically, terminal 1 receives when registering request, described location server 4 mates with the semantic information in described interior space position model data base according to the character position of described terminal 1 sign-in desk in described geometric coordinate, obtains the semantic locations information of described sign-in desk；By receive register dynamic text record content and image content is uploaded；Meanwhile, show the semantic locations of sign-in desk and on map, show position.Here, terminal 1 can enter social networks by the wireless network that Wi-Fi sets up, it is also possible to enters social networks by the 3G/4G network of mobile operator.

Here, terminal 1 is before content is registered in issue, it is also possible to arrange visible authority, i.e. this register content and position visible to which good friend, and invisible to which good friend, to protect the privacy of user.

In actual application, people the most often can record the movable dynamic of special meaning, has gone to such as weekend certain delicious dining room to taste cuisines, or has gone Conference Hall to listen to report, or have purchased the clothes etc. of material benefit certain clothes shop.Register and dynamically feature people's authentic activity information in somewhere at a moment in time in real world in position, and indoor fine-grained position granular information can more truly reflect the residing space operation of people.Being registered after dynamically sharing in the middle of mobile social networking in these positions by user, gradually constructs the user vivid label in mobile social networking, meets user and wishes to build demand in the in the eyes of impression of good friend.Such as, the position that user issues in library, bookstore is registered the most more, then the good friends of this user just have in the heart label and the impression such as " scholar-tyrant, one who exercises autocratic control in academic and educational circles ", " having deep love for study " to this user.

In the present embodiment, terminal 1 Real-time Collection alignment sensor data, location-server 3 utilizes the real-time geometric coordinate of hybrid positioning technology (PDR method, Wi-Fi and bluetooth) computing terminal 1, it is ensured that user issues when registering, the dynamic of sign-in desk geometric coordinate and high precision.

Embodiment three

Corresponding to embodiment one, described location server 4 is going out the distance calculated between described terminal 1 and each good friend's terminal；After described good friend's terminal (with tabular form) is shown on the interface of described terminal 1 by described social networking application server 5 according to the distance of distance, described terminal 1 is additionally operable to send, to target good friend's terminal 2, the request of tracking by instant communication server 6.

When carrying out good friend and following the trail of, owing to relating to the access rights of customer location, therefore introduce and follow the trail of request mechanism, the most described terminal 1 sends good friend by described instant communication server 6 to target good friend's terminal and follows the trail of request, after described good friend's request of following the trail of is permitted, the request of described good friend being followed the trail of of described terminal 1 sends to described location server 4；Described location server 4 receives after good friend follows the trail of request, follow the trail of according to described good friend and ask to obtain the second geometric coordinate of described target good friend's terminal 2 to described location-server 3 with the 3rd cycle preset, and matched by fine granularity layer node in described second geometric coordinate and described interior space position model data base and to obtain the most close node, and obtain the position semantic information of site layer corresponding to this node；The position semantic information of described target good friend's terminal 2 is shown on interface by described terminal 1.Wherein, described matching process is identical with the building process of the matching process of embodiment one, interior space position model with the building process of interior space position model, does not repeats them here；Described 3rd cycle preset can be 1～3HZ, it is preferable that can be 1HZ, 1.5HZ or 2HZ.

Here, after described tracking request is confirmed, terminal 1 and target good friend's terminal 2 all can check mutually the positional information of the other side, to understand mutual distance each other.

And when target good friend is specific group, following the trail of request mechanism needs to be forced to allow or be arranged in advance to allow；Wherein, described specific group may include that child, old man or patient etc..

Further, described terminal 1 can also receive the geography fence of default described target good friend's terminal 2, when described location server 4 determines the position of described target good friend's terminal 2 beyond described geography fence, described social networking application server 5 is for pushing reminder message to described terminal 1.Wherein, described geography fence is specially the moving region of described target good friend's terminal 2.

In actual application, when A and B goes window-shopping at indoor mall, owing to crowded in market and two people's focus are not quite similar, therefore two people are likely to each be buried in artificial abortion, are broken up by artificial abortion.So A and B just can use good friend's dynamic tracing the position of real time inspection the other side, to arrange prompting scope simultaneously, and when there being a side to walk out this scope, the opposing party can receive this good friend and walk remote reminder message, and then this user can quickly recognize and to go to find good friend.

If it addition, in the indoor place such as nursing house, kindergarten, hospital, caregiver is limited, then be equipped with special terminal positioning device for corresponding tracing object, and caregiver can recognize the position of tracing object in real time by terminal 1.When have tracing object be negligent of concern walk out certain limit time, caregiver will immediately receive prompting message.

In the present embodiment, terminal 1 Real-time Collection alignment sensor data, location-server 3 utilizes the real-time geometric coordinate of hybrid positioning technology (Wi-Fi, bluetooth and PDR method) computing terminal 1, ensure that user carries out good friend when following the trail of, the dynamic of the real-time geometric coordinate of good friend and high precision.

Embodiment four

Corresponding to embodiment one, described location server 4 can also realize the message of fine position information and push, and detailed process comprises the following steps:

Step a: described terminal 1 sends geography fence service request to described location server 4；

Step b: after described location server 4 confirms described request, described location server 4 open position tracking function, the timing of described terminal 1 sends alignment sensor data to described location-server 3；

Step c: described location-server 3 determines the real-time geometric coordinate of described terminal 1 according to described alignment sensor data, sends described real-time geometric coordinate to location server 4；Here, according to described alignment sensor data, described location-server 3 determines that the real-time geometric coordinate of described terminal 1 is identical with the method in embodiment one positioned the real-time geometric coordinate of terminal 1 good friend, do not repeat them here.

Step d: when described location server 4 receives the real-time geometric coordinate of described terminal 1, judge whether this real-time geometric coordinate meets the trigger condition of the PUSH message preset, if meeting, then push corresponding service message to described terminal 1, when terminal 1 sends, to described location server 4, the reading label information reading service message, then this service completes, and the most again pushes corresponding service message to location server 4.If being unsatisfactory for, then continue to obtain from location-server 3 the real-time geometric coordinate of terminal 1, repeat above judge process.Wherein, described trigger condition is default geography fence scope, may include that 10m, 20m etc..

During actual application, when user A have subscribed the message push service of certain point of interest (certain brand shop) after entering market.When Alice strolls near certain brand shop, it will automatically receive the message of the preferential activity in this shop, new product etc. and brand shop.The perimeter query function a kind of from embodiment is different, and this function is to need to subscribe to specific interest point information and according to being located proximate to the service that relation automatic reception is subscribed to.Certainly, interest point information is not limited to the Business Information mentioned in the present embodiment, also includes exhibition position, museum, office building section office, airport duty-free shop etc..

Embodiment five

Corresponding to embodiment one, described terminal 1 is additionally operable to send inquiry request and query argument to location server 4；Described location server 4 for meeting the object of described query argument according to the search of described inquiry request, and the geometric coordinate meeting the query object of query argument is back to described terminal 1.Here, described query argument includes: query object classification, query context, point of interest category and inquiry quantity；Described query object classification includes: periphery good friend inquiry and periphery point of interest are inquired about.Described point of interest category includes: digital product, clothing, cuisines etc..Described query context may include that 10m, 20m, 50m or 100m etc..

After described location server 4 receives inquiry request and query argument, it is judged that described query object classification；When determine described query object classification be periphery good friend inquire about time, send inquiry request to described social networking application server 3, described social networking application server 5 sends the friend information of terminal 1 according to described inquiry request to described location server 4；After described location server 4 receives described friend information, obtain the real-time geometric coordinate of each good friend of described terminal 1 to described location-server 3 with the 4th cycle preset according to friend information.Wherein, the method for the real-time geometric coordinate obtaining each good friend in the present embodiment is identical with the method obtaining the real-time geometric coordinate of terminal 1 in embodiment one, does not repeats them here.Described friend information may include that good friend's head portrait, title and good friend's social networks etc..Wherein, the described 4th cycle preset was 1Hz.

After described location-server 3 determines the real-time geometric coordinate of described each good friend of terminal 1, the real-time geometric coordinate of each good friend is stored to location database, and according to the friend information of described location server 4 transmission, the real-time geometric coordinate information of corresponding good friend is sent to location server 4, after described location server 4 receives the real-time geometric coordinate information of the corresponding good friend of terminal 1, by the fine granularity layer of described real-time geometric coordinate information MAP to interior space position model.Wherein, described real-time geometric coordinate intuitively can not be shown by terminal, and described semantic locations information can be by what terminal intuitively showed.Described interior space position model is made up of fine granularity layer, outlet layer and site layer.The adaptivity graph model that fine granularity layer is made up of node and limit, node represents interior space specific location point, while the annexation illustrated between each location point.Outlet layer and site layer are then from the abstract outlet diagram obtained of fine granularity layer and semantic locations figure, and outlet diagram represents outlet node and topological relation thereof, and semantic locations figure then represents indoor each sub spaces and topological relation thereof.Wherein, the building process of described interior space position model is identical with the building process of the interior space position model in embodiment one, does not repeats them here.

When the real-time geometric coordinate information MAP of the corresponding good friend of terminal 1 is to after the fine granularity layer in interior space position model, described location server 4 also utilizes same localization method to determine the real-time geometric coordinate of described terminal 1, and utilize perimeter query algorithm search to meet the object of described query argument, and the geometric coordinate information of the described query object meeting query argument is back to described terminal 1.In the present embodiment, the detailed process of perimeter query algorithm based on level indoor location model is as follows:

(1) inquire about the geometric coordinate of moving reference point (i.e. initiating the terminal 1 of inquiry request) and obtain its corresponding network node；

(2) obtained the result of search tree for the first time by hierarchical network extension, and in the range of the extension of this network, obtain the mobile object meeting condition；

(3) if reference point does not move, then network expanded search tree also will not change, and can directly obtain the mobile object meeting condition；

(4) if reference point moves, the root node of web search tree, the network node that root node is mapped by current reference point are updated；

Next (5) boundary node is obtained according to inquiry based on previous moment position, it is judged that this boundary node has overruned threshold value the most, if in range threshold, then proceeds network extension；Wherein, described range threshold is default mobile range, such as, 10m, 20m etc..

(6) if this boundary node is not in range threshold, then its father node of backward tracing, delete all distance values node more than scope threshold values along father node pointer, obtain the network expanded search tree after updating；

(7) the network expanded search tree after finally traversal updates is met the mobile good friend of condition.

Further, the hierarchical network extended method detailed process involved by step (2) is as follows:

A () obtains unitary space, reference point present position name identification according to reference point, carry out network extension in the fine granularity layer figure that this mark is corresponding, stops when expanding to the outlet node that this position units space connects.

B the extension of () network is switched to outlet layer and is extended, the all outlet ports node of moving reference point present position space cell is carried out network extension, and the most all distances to reference point all will be added in network expanded search tree less than or equal to the outlet node of range threshold.

The leafy node of c network expanded search tree that () step 2 obtains is outlet node, obtains corresponding fine granularity layer figure according to its locational space unit connected, and carries out network extension at fine granularity layer.Extension is stopped when being expanded the node distance to reference point more than range threshold.

It addition, when described query object classification is the inquiry of periphery point of interest, described location server 4 is specifically additionally operable to: read interest point information in (connected reference) described interior space model；Perimeter query algorithm is utilized to calculate described point of interest result.And point of interest result is back to terminal 1.Wherein, the process of inquiry point of interest is just the same with the query script of inquiry periphery good friend, does not repeats them here.

The peripheral position query function that the present embodiment provides, utilizes particle filter to merge location algorithm, positions periphery friend location and point of interest in conjunction with interior space position model, improve positioning precision and dynamic；Utilize perimeter query algorithm to calculate periphery good friend and the Query Result of point of interest, improve the accuracy of Query Result, improve Consumer's Experience.

The above, only presently preferred embodiments of the present invention, it is not intended to limit protection scope of the present invention, all any amendment, equivalent and improvement etc. made within the spirit and principles in the present invention, should be included within the scope of the present invention.

Claims (10)

1. an indoor social navigation system, it is characterised in that described system includes:

Location-server, sends out for the first alignment sensor data and target good friend's terminal receiving terminal transmission The the second alignment sensor data sent, calculate the first of described terminal according to described first alignment sensor data Geometric coordinate, calculates the second geometric coordinate of described target terminal according to described second alignment sensor data； And described first geometric coordinate and described second geometric coordinate are stored in location database；

Location server, is used for obtaining indoor location model data, according to described interior space position data structure Build the space topological network of each level of indoor location model；

Receive the navigation requests that described terminal sends, obtain to described location-server according to described navigation requests Described first geometric coordinate and described second geometric coordinate, and utilize path search algorithm at described space topological Network calculates optimum guidance path.

2. the system as claimed in claim 1, it is characterised in that the sky of the described each level of indoor location model Between topological network figure include: fine granularity level AEGVG figure, outlet level illustraton of model and position hierarchical model Figure.

3. system as claimed in claim 2, it is characterised in that described location server is according to described indoor Spatial position data builds the fine granularity level AEGVG figure of indoor location model and specifically includes:

Extract one-dimensional skeleton according to indoor floor plan, form the one-dimensional Voronoi in interior space elongated zones Figure；

Described open area is carried out stress and strain model with the default length of side and forms grid chart, described grid chart is added In described Voronoi diagram；

Carry out sampling node with the average step length of pedestrian for the sampling interval, generate described AEGVG figure.

4. system as claimed in claim 2, it is characterised in that described location server is according to described indoor Spatial position data builds the outlet level illustraton of model of indoor location model and specifically includes:

The outlet joint of described coarser grained layers is determined according to the exit position in described fine granularity layer AEGVG figure Point；

Reachable path between adjacent position is built as limit described outlet level illustraton of model.

5. system as claimed in claim 2, it is characterised in that described location server is according to described indoor Spatial position data builds the position level illustraton of model of indoor location model and specifically includes:

The position joint of described coarser grained layers is determined according to the character position in described fine granularity layer AEGVG figure Point；

Described position level illustraton of model is generated according to adjacent, the connected relation between described nodes of locations.

6. the system as claimed in claim 1, it is characterised in that described location-server is for according to described First alignment sensor data calculate the first geometric coordinate of described terminal, specifically include:

When described location-server detects the anchor point signal in described first alignment sensor data, by described Anchor point signal and location fingerprint data base carry out fingerprint matching and determine the initial position of described terminal；

Detecting described anchor point signal with default cycle timing again, if described anchor point signal being detected, utilizing grain Son filtering is merged location algorithm and is merged pedestrian's dead reckoning PDR method, described anchor point signal and interior space letter Breath determines the first geometric coordinate of described terminal.

7. the system as claimed in claim 1, it is characterised in that when described location server builds indoor position After putting the space topological network of each level of model, specifically it is additionally operable to:

Receive the factor of influence in every bar navigation path in described space topological network；

Receive each described factor of influence weighing factor to current navigation path；

The comprehensive weights of every paths are calculated according to described weighing factor.

8. system as claimed in claim 7, it is characterised in that described factor of influence specifically includes: indoor Pedestrian's reach distance, up to time, density of personnel and road width.

9. system as claimed in claim 8, it is characterised in that described indoor pedestrian's reach distance is by formulaCalculate；Wherein, described O_iFor described terminal corresponding first Mobile object；Described O_jFor described target good friend's terminal corresponding second move object；Described (x_x,y_k) it is thin Granularity layers first moves object O described in distance_iNearest node n_kCoordinate；Described m is integer.

10. the system as claimed in claim 1, it is characterised in that the appraisal letter of described path search algorithm Number is: f (n)=g (n)+h (n)；Wherein, described f (n) is that start node is via node n estimating to destination node Valency function；Described g (n) is that start node described in state space is to the actual cost of node n；Described h (n) Actual cost for node n to the optimal guidance path of described destination node.