CN107084735A - Guidance path framework suitable for reducing redundancy navigation - Google Patents
Guidance path framework suitable for reducing redundancy navigation Download PDFInfo
- Publication number
- CN107084735A CN107084735A CN201710283551.7A CN201710283551A CN107084735A CN 107084735 A CN107084735 A CN 107084735A CN 201710283551 A CN201710283551 A CN 201710283551A CN 107084735 A CN107084735 A CN 107084735A
- Authority
- CN
- China
- Prior art keywords
- path
- module
- user
- terrestrial reference
- route
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
- G01C21/3453—Special cost functions, i.e. other than distance or default speed limit of road segments
- G01C21/3484—Personalized, e.g. from learned user behaviour or user-defined profiles
Abstract
The invention discloses a kind of guidance path framework for being applied to reduce redundancy navigation.It includes data preprocessing module, path division module and path and summarizes module, and data preprocessing module is connected with path division module, and path division module is summarized module with path and is connected；Data preprocessing module, for extracting user knowledge；Path division module, is divided for finding optimal path；Module is summarized in path, for generating personalized path navigation known to user on path；Data preprocessing module includes matching and calibration module, terrestrial reference extraction module and user knowledge measurement module, and matching and calibration module, terrestrial reference extraction module are connected with user knowledge measurement module respectively, and user knowledge measurement module is connected with path division module.Based on framework of the present invention, navigation application program is developed, navigation instruction can be simplified, the development cost of Navigator, demand of the reduction navigation application to resource, it is possible to so that navigation information is easier to be easily absorbed on a cognitive level by the user is reduced.
Description
Technical field
The present invention relates to field of navigation technology, specifically a kind of guidance path framework for being applied to reduce redundancy navigation.
Background technology
Navigation application is a kind of application that best route and corresponding turn direction are found in road network.Existing navigation application
The navigation Service being made up of turn direction (turn-by-turn) provided is the special extraction information from the road network of bottom and is got
's.So, the information that it can be easily translated into physical world goes to tell about (how far/how long turn direction).But this translation, neglect
Cognition of the mankind to geographical space has been omited, it is often very tediously long for understanding the driver of geographic area.On most of roads
Driver has good city knowledge and is familiar with some of road network part, the route that they take, (for example, route from
Family arrives neighbouring highway, can in short be summarized：From the home to XXX highways；Rather than：Above XX meters of left-hand rotations ... ...).Therefore,
If to part road known to driver, also navigated using the turn direction of existing navigation application, then actually navigation system
To be very redundancy and complexity, corresponding Navigator is also high to the demand of resource, cause navigation application development cost and
Operation cost is also high.Also, if used always by direction navigation is turned, the information that user receives is typically that " front XX meters left
Turn, front XX meters of right-hand rotation, front keep straight on XX meters ... ... ", the navigation information that user receives is single, it is difficult to meet individual character
The navigation needs of change.Especially with the development of development of Mobile Internet technology, the navigation application of mobile device and onboard navigation system
Ground constantly increases, and have accumulated the track data of substantial amounts of user generation, using these big datas produce more effectively, be easier quilt
The navigation information that people understands is highly useful.
Existing path navigation framework is all based on the path navigation framework of turn direction, and they are extracted from the road network of bottom
Road network information, then carries out respective handling, does not account for structure of the path known to user in navigation framework.Also,
After the known paths of user are extracted, these routes are probably sparse, but also may include significantly not true
It is qualitative, how to be instructed using known path it is actual be driven in existing navigation solution there is not yet.
The content of the invention
It is an object of the invention to overcome the deficiencies of the prior art and provide a kind of navigation road for being applied to reduce redundancy navigation
Footpath framework, to be at least up to based on framework of the present invention, develops navigation application program, simplifies navigation instruction, reduction Navigator
Development cost, reduces demand of the navigation application to resource, and cause navigation information to be easier the effect being easily absorbed on a cognitive level by the user.
The purpose of the present invention is achieved through the following technical solutions：Guidance path frame suitable for reducing redundancy navigation
Frame, it includes：Module is summarized in data preprocessing module, path division module and path, and data preprocessing module is divided with path
Module is connected, and path division module is summarized module with path and is connected；
Described data preprocessing module, for extracting user knowledge；
Described path division module, is divided for finding optimal path；
Module is summarized in described path, for generating personalized path navigation known to user on path；
Described data preprocessing module includes matching and calibration module, terrestrial reference extraction module and user knowledge measurement mould
Block, matching and calibration module, terrestrial reference extraction module are connected with user knowledge measurement module respectively, user knowledge measurement module and road
Footpath division module connection；
Described matching and calibration module, for track calibration and trajectory clustering, extract the path letter that user often accesses
Breath, the routing information that described user often accesses includes avenue title；
Described terrestrial reference extraction module, for from the historical trajectory data of user, extracting landmark information, described terrestrial reference
Information includes start node and target point；
Described user knowledge measurement module, the known paths for exporting user.
Further, it includes：Subscriber interface module, described subscriber interface module is connected with path division module, also,
Also summarize module with path to be connected, for inputting path and summarizing the personalized path navigation of module generation for outgoing route.
Further, it includes：Data source modules, described data source modules are connected with data preprocessing module, for leading
Enter pending data；The source of described pending data, including：Track data recording equipment, mobile terminal navigation should
With the historical trajectory data applied with vehicle terminal navigation, interest point data source and map data base.
Further, in including route known to user and a route of the unfamiliar route of user or a plurality of route,
For route known to user, user is instructed using the personalized route guidance；For the unfamiliar route of user, using turning
Curved direction navigation instruction user.
Further, route is in the push to weigh user by calculating a familiarity fraction f (R), it is described
Familiarity fraction f (R) is calculated and obtained according to the pass by meaning of the frequency in the path and the terrestrial reference in the path of user；Wherein,
F (R) size interval range includes：f(R)∈[0,1].
Further, to path segments R (i, j) familiarity, using equation below calculate familiarity fraction f (R (i,
j))：
Wherein,It is the length ratio of path segments, g () is a monotonic function, g (0)=0, g (1)=1, R
(i, j) is a R path segments, is terminated since R (i) to R (j), and R (i) is path R i-th of terrestrial reference, and R (j) is path
R j-th of terrestrial reference.
Further, in a kind of travel navigation scene, to route known to user, it assign passenger as authority, terrestrial reference conduct
Center, registration office carries out the calculating of the meaning of terrestrial reference using HITS algorithms as hyperlink, and then user passes by the path
Weighted sum is normalized in the result of calculation of frequency and the meaning of terrestrial reference, using the result of the normalization weighted sum as
Know the familiarity fraction f (R) in path；For natural route, then its familiarity fraction f (R) is set to a steady state value.
Further, the workflow of the framework, including step：
S1：From the historical trajectory data of user, path known to extraction user makes user have phase to known path
That answers is familiar with value；
S2：Based on road network information architecture natural route；
S3：Optimal path is found using Dynamic Programming to divide, or, first make path proximity using path candidate condition is relaxed
Matching, then find optimal path division；
S4：Path segments after being divided based on path, the personalized path of generation is summarized, and instructs user.
Including sub-step further, in step sl,：
S101：Calibrated using map match with the track based on anchor, initial trace T is converted into a rail based on terrestrial reference
Mark, anchor point is used as by handling terrestrial reference；After calibration, on the basis of track alignedly target, the track of calibration is used as a ground
Target sequence, as path；
S102：After trajectory path has been demarcated, paired similar of the trajectory path demarcated is calculated using EDR distance algorithms
Property, and constitute using similar path path cluster；
S103：Using equation below, most representative path R in each path cluster is selected*It is used as known paths；
Wherein, R is path,For path cluster, DEDRFor the EDR distances of two paths；
S104：User is weighed using path familiarity fraction f (R) and path segments familiarity fraction f (R (i, j))
To the familiarity in path.
Including sub-step further, in step s 2,：
S201：Input road network, and initialization path clusterMakeFor empty set；
S202：According to the Name ＆ Location of path segments in road network, it is ranked up using comparator, and from road network
According to the Rule Extraction a line of agreement in network, a line of extraction is assigned to setting variable e1；
S203：The side extracted in step S202 is removed in road network, then, this side of removal path is deposited into
In R；
S204：In remaining road network, searched using path R and belong to same section, be joined directly together and turn
Angle is less than remaining path of setting value, and completes the link between the path that finds, finally links these paths and constitutes
Path candidate cluster；
S205：The Name ＆ Location of path segments in path candidate cluster is sorted using comparator, and from step
According to the Rule Extraction a line of agreement in path candidate cluster in S204, a line of extraction is assigned to setting and become
Measure e2；
S206：The side extracted in step S205 is removed from road network；
S207：By path R and find, satisfactory setting variable e2In path connection, constitute new path；
S208：When path R can not find remaining path candidate cluster again, terminate to set variable e1Headed by path
Link process, path cluster is deposited into by the new path foundIn；
S209：Circulation performs step S202~S208, and traversal searches all paths in road network, until road network
In there is no manageable road segment segment untill, then, return path cluster
Further, in step s3, including：
(1) the use Dynamic Programming described in is found optimal path and divided, including sub-step：
S3012：Whole path R is traveled through, the path candidate cluster that retrieval is linked comprising path has retrieved all path candidates
Afterwards, the condition optimal dividing of the subpath in whole path R is calculated one by one；
S3013：Path is linked to divide to merge with current path and generates new path division, equation below increment meter is used
The quality score that newly-generated path is divided：
Wherein, the quality score that Q (i, R'(j, k)) divides for newly-generated path；For path segments R (i-
1, path candidate cluster i)；(R'(j k) is path segments R'(j, familiarity fraction k) to f；λ is one that user specifies
Nonnegative constant；(i-1 i) arrives R (i) path segments to R for R (i-1)；R " (l, m) is the path segments that R " (l) arrives R " (m)；R'
(j, k) be R'(j) arrive R'(k) path segments；R " (l) is path R " l-th of terrestrial reference；R " (m) is path R " m-th of ground
Mark；R'(j) j-th of terrestrial reference for being path R'；R'(k) k-th of terrestrial reference for being path R'；
Then, selection quality score highest one is divided as optimal path；
S3014：Circulation performs step S3012~S3014, calculating is constantly iterated, until finding whole path R most
Shortest path is divided, then, is returned to the whole path R optimal paths and is divided.
(2) use described in relaxes path candidate condition and first makees path proximity matching, then finds optimal path division, including
Sub-step：
S3021：Define route matching relation and define path similarity relation, including：
(a) route matching relation is defined：Two paths R and R` are given, each path is represented by a series of terrestrial reference, one
Route matching M (R, R`) is a terrestrial reference arranged properly to set, and each first terrestrial reference of terrestrial reference centering comes from R, second
Terrestrial reference comes from R2；
(b) path similarity relation is defined：Given two paths R and R`, and if only if, and R and R` has common terminal, and
In the presence of an optimal path matching M (R, R`), then R is similar to R`；Described optimal path matching M (R, R`) is met：
Each terrestrial reference in R and R` of I occurs at least one times in M (R, R`)；
Each terrestrial reference of II in M (R, R`) is to (li, li`), their network space DN(li,li`)≤ε；
The terrestrial reference pair intersected is not present in III in M (R, R`)；
Wherein, ε is the maximum road network spacing between the terrestrial reference pair of matching；
S3022：It is that last path segments is R'(j to set Q (i, R'(j, k)), R k) (1, i) relax it is optimal
The quality score that path is divided, when R (i) is a R terminal, regulation R (i)=R` (k), the then optimal path relaxed is divided
Meet following recurrence formula：
The condition optimal path relaxed is calculated to divideIncluding following two situations：
If 1. k-j>1, then the condition optimal path relaxed is divided intoWith
Last path segments come from same paths, enumerate when R (i-1) matching R` (k), R (i-1) matching R` (k-1), R (i)
Match R` (k-1) when three kinds of situations, then incrementally calculated using equation below path division quality score Q (i, R'(j,
k))：
Q(i,R'(j,k))←max{Q(i-1,R'(j,k)),Q(i-1,R'(j,k-1))+f(R'(j,k))-f(R'(j,
k-1)),
Q(i,R'(j,k-1))+f(R'(j,k))-f(R'(j,k-1))}
If 2. k-j=1, that is, terminate the search procedure of path candidate, a new route segment R` (j, k) is opened, now
Path candidate clusterR` (j) is contained, also, R`` isIn arbitrary path, for a paths
There are two kinds of situations in section R`` (l, m), wherein R`` (m)=R` (j)：R`` (m) and R (i-1) is matched or R`` (m) and R (i)
Match somebody with somebody, enumerate both of these case, the quality score Q (i, R'(j, k)) of path division is then incrementally calculated using equation below：
Then, selection quality score highest one is divided as optimal path.
Further, in step s 4, including：
(1) for natural route section, it is described using multiple key features；Described multiple key features include street
Name, distance and shift strategy, described shift strategy represent the action taken at the end of natural route section, it includes turning to,
Continue to travel, leave and reach；
(2) for known paths section, it is described by describing its starting terrestrial reference and terminal terrestrial reference, in starting point terrestrial reference and
Terminal terrestrial reference be not enough to description it in the case of, along this paths travel turnpike road be also included within description；According to
Actual conditions, can omit one or more features, and the feature of omission includes starting point, street name..
The beneficial effects of the invention are as follows：
(1) historical trajectory data is calibrated and clustered first to framework of the present invention, extracts user knowledge, including known road network
Given path is divided into multistage by network information, the known path of output as given path, is divided to find optimal path, most
Improve to limits its path division quality to lay the foundation, generating high-quality personalization based on optimal path splitting scheme leads
Boat information；For example, the navigation application program based on the present invention, the historical trajectory data generated using abundant user, from road
Terrestrial reference, the route that terrestrial reference (for example, profit point or crosspoint) and user often access are extracted in network, and uses the information extracted
It is utilized for each user's customization routing strategy；
(2) with mobile device and onboard navigation system navigation application constantly increase, have accumulated substantial amounts of user life
Into track data, based on the present invention, can use these data can produce more effectively, be easier the navigation that is more readily understood
Information；
(3) based on the present invention, navigation information can be become easier to be understood by driver, meanwhile, the simplification of instruction also can
The demand (space of such as bandwidth and screen) of resource is significantly reduced, in addition, such path framework may also be used for instructing new
Emerging autonomous driving vehicle, they need not follow detailed turn direction, and go to pay close attention to those higher level information；
(4) based on the present invention, using content that is specific, being readily appreciated that replace those redundancies, known by driver
Part, and the unfamiliar place of driver still guides them using detailed steering；
(5) based on the present invention, terrestrial reference and known track are found, and navigation direction is generated succinctly using them, still
The enough information of transmission is explained route for user, simplifies navigation instruction, reduces navigation redundancy；
(6) based on the present invention, using known route, these routes are probably sparse, and are possible to comprising very big
The uncertainty of degree, is to be adjusted to " approximate match " from " matching completely " by the requirement of alternative route, is ensureing that transmission is enough
Information come on the premise of explaining route, so as to summarize path using route known to more users, it is considered to actual
Driving behavior, not only significantly reduces algorithm redundancy, and causes navigation information to be easier to be easily absorbed on a cognitive level by the user, and can be carried to driver
For personalized navigation information；
(7) based on the present invention, substantial amounts of experiment has been carried out on true and synthesis track data collection, has as a result shown, carries
Take an appropriate number of known route, it is possible to reduce the quantity of redundancy navigation is more than 60%, while enough information still can be provided,
For user's guiding route；
(8) based on the present invention, on the one hand make navigation information become easier to be easily absorbed on a cognitive level by the user, it is another face to face, the present invention letter
Navigation instruction is changed, has significantly reduced demand of the navigation application to resource, the development cost and navigation for reducing Navigator should
Operation cost；
(9) navigation Service being made up of turn direction that existing navigation application is provided, is the special extraction from the road network of bottom
Information and get, by comparison, in the track data accumulated from user based on the present invention or other track data sources, carry
Family known paths information, and the Given information based on extraction are taken, generation meets the navigation information of user, facilitates user, carry
High user's navigation usage experience；
(10) conventional guidance path algorithm frame, input is generally made up of starting point and destination, and output is to be based on
The best route of standard (such as conventional, safety), and be a route in the input of the present invention, output is a route summary,
Target is to find a route, best suits the preference of user；The present invention is the navigation system of bottom, therefore can be with any navigation
The system integration, for example, navigation system is recommended in the personalized tourism of one kind based on the present invention, can take what is produced by navigation system
Any route, and being mapped in a pattern travelled based on user's history, thus find best suit user preference route it is comprehensive
State that there is provided the service of Personalized Navigation confidence.
Brief description of the drawings
Fig. 1 is the step flow chart of the inventive method；
Fig. 2 extracts the step flow chart of user's known paths for the present invention；
Fig. 3 builds the step flow chart of the natural route based on road network information for the present invention；
Fig. 4 directly finds the step flow chart of optimal path division for the present invention using Dynamic Programming；
Fig. 5 first makees path proximity matching for the present invention using path candidate condition is relaxed, then finds optimal path division
Flow chart of steps；
Fig. 6 is the structure chart of framework of the present invention；
Fig. 7 is a kind of system construction drawing of embodiment of the present invention.
Embodiment
Technical scheme is described in further detail below in conjunction with the accompanying drawings, but protection scope of the present invention is not limited to
It is as described below.
Embodiment
As shown in fig. 6, a kind of guidance path framework for being applied to reduce redundancy navigation, it includes：Data preprocessing module,
Module is summarized in path division module and path, and data preprocessing module is connected with path division module, path division module and road
Summarize module connection in footpath；
Described data preprocessing module, for extracting user knowledge；
Described path division module, is divided for finding optimal path；
Module is summarized in described path, for generating personalized path navigation known to user on path；
Described data preprocessing module includes matching and calibration module, terrestrial reference extraction module and user knowledge measurement mould
Block, matching and calibration module, terrestrial reference extraction module are connected with user knowledge measurement module respectively, user knowledge measurement module and road
Footpath division module connection；
Described matching and calibration module, for track calibration and trajectory clustering, extract the path letter that user often accesses
Breath, the routing information that described user often accesses includes avenue title；
Described terrestrial reference extraction module, for from the historical trajectory data of user, extracting landmark information, described terrestrial reference
Information includes start node and target point；
Described user knowledge measurement module, the known paths for exporting user.
Further, it includes：Subscriber interface module, described subscriber interface module is connected with path division module, also,
Also summarize module with path to be connected, for inputting path and summarizing the personalized path navigation of module generation for outgoing route.
Further, it includes：Data source modules, described data source modules are connected with data preprocessing module, for leading
Enter pending data；The source of described pending data, including：Track data recording equipment, mobile terminal navigation should
With the historical trajectory data applied with vehicle terminal navigation, interest point data source and map data base.
Basic variable is defined, including：
It is a geographic point in space to define terrestrial reference l, a terrestrial reference l, and it is stable, independently of user trajectory
Outside, it can be a node on a point of interest or road network.
Path R is defined, the paths R in road network is defined as the sequence of terrestrial reference composition one by one, R=l1,
L2 ... ln, R i represent i-th of terrestrial reference in the path, and the adjacent ground of each two, which is marked in road network, to be joined directly together.
Define path segmentsConsider paths a R=l1, l2 ..., ln, its path segmentsLj, is defined as a terrestrial reference sequence subset in R, wherein, R (i, i+1) represents connection R (i)
With R (i+1) route segment, R (1, n) represent path R in itself.
It is i-th of terrestrial reference in the R of path to define R (i).
It is a path segments to define R (i, j), is terminated since R (i) to R (j).
It is the quality score that a path divides P to define Q (P).
In road network, adjacent link generally has some common features, such as street name, direction.Based on this
A little features, can be divided into road network in multiple disjoint routes, i.e. each link fully belongs to a route, same
All adjacent links have identical feature in one route.Because such route is independently of specific user,
It is defined as natural route (natural route, NR), and natural route is navigated mode that software navigates with turn direction to retouch
State stretch.For given path R=l1, l2 a ... ln, it can be represented with the sequence of a natural route fragment,
It can be expressed as：[NR1(,),NR2(,),…,NRm(,)](m<=n), it is called the navigation strategy of system optimal.
But, also there is another type of route, it varies with each individual.In real world, the commuter in city
(user) some terrestrial references and route can be especially familiar with, such as from the home to job site or from job site to shopping in
The heart, we call this path for known paths (known route, KR), and a known paths often include multiple natural roads
Footpath section, such as KR=[NRi (), NRi+1 () ..., NRj ()].Therefore, if this natural route section sequence is one of R
Subsequence, we can be the optimal navigation of user to replace this subsequence using a single known paths section
Strategy.Because multiple natural route segments are generalized a known route segment, the optimal navigation strategy of user can make navigation
Become more succinct, i.e. the navigation strategy than system optimal has used less path hop count, can cook up user and be more familiar with
Route.
A kind of personalized method for path navigation, including：
From track data, extract user knowledge, the user knowledge based on extraction finds path known to user, with
Known to family on path, generation meets the personalized path navigation of user；
The user knowledge includes the routing information and landmark information that user often accesses, what described user often accessed
Routing information includes avenue name information；The source of the track data, including track data recording equipment, mobile terminal
Navigation application and vehicle terminal navigation application.
Further, in including route known to user and a route of the unfamiliar route of user or a plurality of route,
For route known to user, user is instructed using the personalized route guidance；For the unfamiliar route of user, using turning
Curved direction navigation instruction user.
Further, as shown in figure 1, described method includes step：
S1：From the historical trajectory data of user, path known to extraction user makes user have phase to known path
That answers is familiar with value；
S2：Based on road network information architecture natural route；
S3：Optimal path is found using Dynamic Programming to divide, or, first make path proximity using path candidate condition is relaxed
Matching, then find optimal path division；
S4：Path segments after being divided based on path, the personalized path of generation is summarized, and instructs user.
Further, as shown in Fig. 2 in step sl, including sub-step：
S101：Calibrated using map match with the track based on anchor, initial trace T is converted into a rail based on terrestrial reference
Mark, anchor point is used as by handling terrestrial reference；After calibration, on the basis of track alignedly target, the track of calibration is used as a ground
Target sequence, as path；
One original track T is the sequence of a limited position, from the original path and its phase of a mobile object
The timestamp sampling of pass, i.e. T=(P1, T1), (P2, T2) ... ... (PN, TN), wherein Pi is the position specified by latitude and longitude
Put, Ti is corresponding timestamp.In order to preferably analyze track, we calibrate the track using map match and based on anchor, will
Initial trace T is converted into a track based on terrestrial reference, and anchor point is used as by handling terrestrial reference.After calibration, in these tracks pair
On the basis of quasi- terrestrial reference, the track of calibration can be considered as the sequence of a terrestrial reference, i.e. path.
S102：After demarcation path, paired similar of the path demarcated is calculated using EDR distance algorithms
Property, path cluster is constituted using similar path；
S103：Using equation below, most representative path in each cluster is selected to be used as known paths R*；
Wherein, R is path,For path cluster, DEDRFor the EDR distances of two paths；Most representative path is
The minimum path with other all path distance sums；
In these known routes, route known to driver may be different.For example, for from the home to yard
Route, driver knows the change in track and common traffic.On the contrary, only having walked route several times for user, take charge of
Machine may not know to want Zou Zhetiao roads.Therefore, to each known route, calculate fraction f (R) ∈ [0,1] and weigh
User's is in the push.F (R) is called familiarity fraction, generally, f (R) influence comes from the following aspects：
(1) pass by the frequency in the path；(2) meaning of beginning and end terrestrial reference.Calculating to f (R), as a kind of embodiment, one
Plant in travel navigation application scenarios, HITS algorithms can be used, the algorithm is using passenger as authority, and terrestrial reference is used as center, registration
Place first carries out the calculating of the meaning of terrestrial reference, then regard the normalization weighted sum of the two factors as known road as hyperlink
The familiarity score in footpath.For a natural path, its familiarity score is set to steady state value.Further, it is possible to use
The method of stationary point detection recognizes point of interest known to user.
It is contemplated that to the familiarity of a paths R, familiarity fraction f is also used to its path segments R (i, j)
(R (i, j)) carries out calculating scoring：
To path segments R (i, j) familiarity, familiarity fraction f (R (i, j)) is calculated using equation below：
Wherein,It is the length ratio of path segments, g () is a monotonic function, g (0)=0, g (1)=1, R
(i, j) is a R path segments, is terminated since R (i) to R (j), and R (i) is path R i-th of terrestrial reference, and R (j) is path
R j-th of terrestrial reference.With length than reduction, it is faster than linear function to be familiar with score g () decrease speed, this with human cognitive more
To be consistent.
S104：User is weighed using path familiarity fraction f (R) and path segments familiarity fraction f (R (i, j))
To the familiarity in path, for natural route, then its familiarity fraction is set to a steady state value.
Natural route is generally made up of multiple route segments with same characteristic features, such as street name, direction.For convenience
Path is collected, and connects adjacent road segment segment, forms new natural route, and first Xuan Yitiao roads are set out, and connect out a big nature
Path, if can not continue connection, just selects a new road to continue to repeat above-mentioned connection procedure, until no road can be elected as
Only, specifically used step S2 is calculated, as shown in Figure 3：
S201：Input road network, and initialization path clusterMakeFor empty set；
S202：According to the Name ＆ Location of path segments in road network, it is ranked up using comparator, and from road network
According to the Rule Extraction a line of agreement in network, a line of extraction is assigned to setting variable e1；
S203：The side extracted in step S202 is removed in road network, then, this side of removal path is deposited into
In R；
S204：In remaining road network, searched using path R and belong to same section, be joined directly together and turn
Angle is less than remaining path of setting value, and completes the link between the path that finds, finally links these paths and constitutes
Path candidate cluster；
S205：The Name ＆ Location of path segments in path candidate cluster is sorted using comparator, and from step
According to the Rule Extraction a line of agreement in path candidate cluster in S204, a line of extraction is assigned to setting and become
Measure e2；
S206：The side extracted in step S205 is removed from road network；
S207：By path R and find, satisfactory setting variable e2In path connection, constitute new path；
S208：When path R can not find remaining path candidate cluster again, terminate to set variable e1Headed by path
Link process, path cluster is deposited into by the new path foundIn；
S209：Circulation performs step S202~S208, and traversal searches all paths in road network, until road network
In there is no manageable road segment segment untill, then, return path cluster。
Further, in addition to, S210：Outgoing route cluster
Up to the present, we just extract known path from the historical track of user, and construct and be based on
The natural route of road network information.
When we describe a route, it is generally divided into logical segment by we, for example, go ahead until stop flag,
Then each section is described again, the present invention is divided, in step S3 based on subregion is concluded based on user knowledge to path
In：
A path R is given, the path defined for R is divided intoCan have
Following conclusion：
It is total path R that all route segments, which are added up, and any two paths are each other without overlapping part, per path segments
RSi can be known route segment or natural route section.Although path R any subregion can produce one always
Knot, but not every is all a good summary.In general, it is intended that the navigation direction of generation has：(1) it is directly perceived
Understand with user is easy to, i.e. using known route, produce known path navigation；(2) succinctly, the quantity of route segment will be most
Smallization.
The mass equation Q (Ρ (R)) of subregion is defined, optimal path is calculated and divides Ρ*(R)：
Wherein, f () is to weigh user to the fractional function of path familiarity, is the quantity of route segment in this subregion,
λ is the nonnegative constant that user specifies, and it is used for punishing the quantity of the route segment of generation, makes the quantity of route segment will not be too
Greatly, it formula is chosen those users and is familiar with, but be unlikely to allow segmentation to become excessive again, finally reach a kind of folding
In effect；" all partition " are all path splitting schemes, and finally selection causes quality Q (Ρ (R)), and score is most
It is optimal splitting scheme Ρ that high path, which divides Ρ (R),*(R)。
Divided as shown in figure 4, optimal path is found in the use Dynamic Programming described in (1), including sub-step：
Optionally, in addition to an initialization step, S3011：Path R is inputted, two-dimensional array D [i] [R is initialized
`], for storage state information, described status information is including last in the parent pointer p for recalling, current path division
One path segmentsThe quality score divided with current path, also, the storage all conditions in two-dimensional array D [i] [R`]
The information that optimal path is divided；And
S3012：Whole path R is traveled through, the path candidate cluster that retrieval is linked comprising path has retrieved all path candidates
Afterwards, the condition optimal dividing of the subpath in whole path R is calculated one by one；
S3013：Path is linked to divide to merge with current path and generates new path division, equation below increment meter is used
The quality score that newly-generated path is divided：
Wherein, the quality score that Q (i, R'(j, k)) divides for newly-generated path；For path segments R (i-
1, path candidate cluster i)；(R'(j k) is path segments R'(j, familiarity fraction k) to f；λ is one that user specifies
Nonnegative constant；(i-1 i) arrives R (i) path segments to R for R (i-1)；R " (l, m) is the path segments that R " (l) arrives R " (m)；R'
(j, k) be R'(j) arrive R'(k) path segments；R " (l) is path R " l-th of terrestrial reference；R " (m) is path R " m-th of ground
Mark；R'(j) j-th of terrestrial reference for being path R'；R'(k) k-th of terrestrial reference for being path R'；
Then, selection quality score highest one is divided as optimal path；
S3014：Circulation performs step S3012~S3014, calculating is constantly iterated, until finding whole path R most
Shortest path is divided, then, is returned to the whole path R optimal paths and is divided.
Also include, S3015：Outgoing route R optimal path divides Ρ*(R)。
One simple searching Ρ*(R) method is that being possible to of dividing of the exhaustive path is combined, and then finds one
Quality highest solution.However, the time complexity of this method and number of paths exponentially other elevational relationship, work as path
In number of links it is larger when, obtain optimal path will become to hang back.So, our method is to use Dynamic Programming,
Optimal path can be found in polynomial time complexity to divide.
As shown in figure 5, path candidate condition is relaxed in the use described in (2) first makees path proximity matching, then find optimal road
Footpath is divided, including sub-step：
Conditional optimal path is defined to divide, for given path R=l1, a l2, ln, its condition is optimal
Path divides Ρ*(R | R` (j, k)) it is defined as the optimal path division that final stage in R is R` (j, k).Particularly, Ρ*(R|R`
()) represent that final stage is all optimal dividings from R` in R.
The optimal dividing on (1) road is that one kind optimal during all path candidates are divided.
Wherein,Be comprising path link R (i-1, the set in all paths i), we investigate it is all can
The situation of energy, then therefrom selects that optimal one kind.
Primarily look at a given path R, using Q (i, R'(j, k)) represent subpath (R (and 1, i) in final stage be
R'(j, the quality of optimal dividing k).
(2) subpath (R (and 1, condition optimal path i) divide quality can (R (1, i-1) condition be most from subpath
Shortest path is divided and derived：
In formula (6), R'(k-1)=R (i-1), R'(j) ∈ R；
In formula (7),Be it is all include path R (i-1, natural route and the collection of known paths i)
Close, to R``, meet following restrictive condition：R`` (i) ∈ R and R`` (m)=R (i-1)；
Prove the derivation formula in (2)：
(1, optimal path i) divides Ρ to subpath R*(R (1, i)), can be from Ρ*(R (1, i-1)) is built, comprising
Following two situations：
1. Ρ is extended*Last path segments of (R (1, i-1)) obtain Ρ*(R(1,i))；
2. a new path segments are created, optimal path is then added it to and divides Ρ*In (R (1, i-1)),
As k-j>When 1, Ρ*(R (1, i) | ()) and Ρ*Last path segments of (R (1, i-1) | ()) are all from same
What one path was obtained, so building Ρ*(R (1, i) | ()) only need to a Ρ*(R (1, i-1) | (R'(j, k-1))) extension one
Down just can be with.Therefore, the quality in newly-generated path can just be calculated with formula (6).
As k-j=1, Ρ*(R (1, i) | ()) and Ρ*Last path segments of (R (1, i-1) | ()) come from
Different paths, we represent last their own path segments with R` and R``.
According to formula：Understand, Ρ*(R (1, i) | (R'(j, k))) it can lead to
Cross addition the last item path and divide Ρ to optimal path*(R (1, i-1) | R`` ()), matter then is divided to new path
Amount, is calculated using formula (7).
From this recurrence formula, the optimal path that R can be calculated with Dynamic Programming is divided, and idiographic flow includes：
First, a two-dimensional array D [i] [R`] is initialized, for storage state information, these status informations include being used for
The parent pointer p of backtracking, last route segment in current divisionThe quality score that current path is divided, D [i] [R`] is deposited
Store up all conditions optimal path and divide Ρ*The information of (R (1, i+1) | R` ())；
Whole path is traveled through, the condition optimal dividing of subpath is calculated step by step, algorithm is retrieved comprising R (i, i+ respectively
1) and R (i-1, path candidate i) can (key be terrestrial reference, and value is to include the road of the terrestrial reference with inverted index or key-value pair
Footpath) effectively complete；
Retrieve after path candidate, design conditions optimal path is divided.New link R (i, i+1) and existing path are drawn
Division simultaneously, and with formula (6) and the quality score of the newly-generated division of (7) incremental computations.Then, quality score highest is selected
One division.Complete after iteration, the optimal path for finding whole route is divided, and can find what is entirely divided by backtracking
Route.
In real life, due to a variety of causes, such as personal like, mistake of noise and track data etc., it is known that
Path tends not to completely overlapped given path.Such as one selected route segmentWith a known paths KR1 very phases
Seemingly, R is described with KR1 can seem more succinct, intuitively, based on our previous definition relations, due to there is subtle difference,
It can not be described with KR1.So in order to overcome this problem, it would be desirable to soften terms, that is, relax and path candidate is wanted
Ask, from identical to approximately the same, realize the matching to path, including：
S3021：Route matching and similarity relation are defined, including：
(a) route matching is defined：Two paths R and R` are given, each path is represented by a series of terrestrial reference, a path
Matching M (R, R`) is a terrestrial reference arranged properly to set, and each first terrestrial reference of centering comes from R, and second terrestrial reference comes from
In R`；
(b) similarity relation is defined：Given two paths R and R`, and if only if, and R and R` has common terminal, and exists
One optimal path matching M (R, R`), then R is similar to R`；Described optimal path matching M (R, R`) is met：
Each terrestrial reference in R and R` of I occurs at least one times in M (R, R`)；
Each terrestrial reference of II in M (R, R`) is to (li, li`), and their road network is smaller than ε；
The terrestrial reference pair intersected is not present in III in M (R, R`).
Wherein, ε is the maximum road network spacing between the terrestrial reference pair of matching, it is considered to two kinds of extreme situations：
If ε is arranged to 0, this has reformed into most the problem of having path division；
If ε is arranged to a sufficiently large value, then arbitrary path can all be considered as similar.
Therefore the setting to ε values has reformed into a personalized path recommendation problem based on historical information, and this value is got over
Greatly, it is more with the more user knowledges of utilization.After the definition for completing similarity relation, then put to discuss by following manner
Wide optimal path partition problem：A path R is given, the high-quality path for finding the R` similar to R divides Ρ*(R`)。
Ρ*(R`) optimal path relaxed for being also referred to as R is divided, and is expressed asIn order to which the optimal path relaxed for finding R is drawn
Point, the situation similar to R` by enumerating all R, the optimal path for then calculating R` using optimal path partitioning algorithm is drawn
Point.It is what we to be found that path with first water, which is divided,.However, the quantity in this kind of potential path is exponential
Not, optimal path partitioning algorithm index is called so that the solution is very poorly efficient.
For to result what is observed in following lemma 3, it has been found that can be for effectively solving using Dynamic Programming
Certainly this problem.On path R, R*Quality highest path is divided in the expression path similar to R.M*(R,R*) represent them
Between a good route matching, (i, k) ∈ M* (R, R*), it is meant that R (i) is matched with R* (k).
Lemma 3：For a good route matching M* (R, R`), if (i, k) ∈ M* (R, R`), then three below
Mark is to ((i-1, k), (i-1, k-1), (i, k-1)) at least has one and be present in M* (R, R`).
To the proof of lemma 3：Assuming that these three terrestrial references are present in neither one in M* (R, R`), for a good road
For the matching of footpath, each terrestrial reference must occur at least one times, we can assume that R (i-1) is matched with R` (l), R` (k-1)
It is to be matched with R (m), because can not there is a situation where to intersect in them, therefore i-1 can be released>M, l<k-1.This means M*
The terrestrial reference of intersection is there is in (R, R`) to (i-1, l) with (m, k-1).
In order to solve this problem with Dynamic Programming, we first define the structure of its subproblem.
S3022：For a paths R, with Q (i, R'(j, k)) be R that last path segments is R` (j, k) (1, i)
The quality that divides of the optimal path that relaxes, when R (i) is a R terminal, regulation R (i)=R` (k), the road then relaxed
The quality of footpath optimal dividing, meets following recurrence Relation：
The optimal path then relaxed is dividedIncluding following two situations：
If 1. k-j>1, then the optimal path that path segments are relaxed is divided intoWithLast path segments come from same paths, enumerate when R (i-1) matching R` (k), R (i-1)
With R` (k-1), three kinds of situations during R (i) matching R` (k-1) then incrementally calculate path quality Q using formula equation below
(i,R'(j,k))：
Q(i,R'(j,k))←max{Q(i-1,R'(j,k)),Q(i-1,R'(j,k-1))+f(R'(j,k))-f(R'(j,
k-1)),
Q(i,R'(j,k-1))+f(R'(j,k))-f(R'(j,k-1))}
If 2. k-j=1, the route segment before terminating, open a new route segment R` (j, k), this is to include R` (j)
The set in path, R`` isIn arbitrary path, for a path segments R`` (l, m), wherein R`` (m)=R` (j),
Here there are two kinds of possible situations：R`` (m) and R (i-1) is matched, and R`` (m) and R (i) are matched, and enumerate both of these case, so
Path quality score Q (i, R'(j, k)) is incrementally calculated using equation below afterwards：
Further, route is in the push to weigh user by calculating a familiarity fraction f (R), it is described
Familiarity fraction f (R) is calculated and obtained according to the pass by meaning of the frequency in the path and the terrestrial reference in the path of user；Wherein,
f(R)∈[0,1]。
Further, to path segments R (i, j) familiarity, using equation below calculate familiarity fraction f (R (i,
j))：
Wherein,It is the length ratio of path segments, g () is a monotonic function, g (0)=0, g (1)=1, R
(i, j) is a R path segments, is terminated since R (i) to R (j), and R (i) is path R i-th of terrestrial reference, and R (j) is path
R j-th of terrestrial reference.
Further, in a kind of travel navigation scene, to route known to user, it assign passenger as authority, terrestrial reference conduct
Center, registration office carries out the calculating of the meaning of terrestrial reference using HITS algorithms as hyperlink, and then user passes by the path
Weighted sum is normalized in the result of calculation of frequency and the meaning of terrestrial reference, using summed result as known route familiarity
Fraction f (R)；For natural route, then its familiarity fraction f (R) is set to a steady state value.
Path is divided into after section, described by order per path segments, generation path is summarized.
(1) for natural route section, it is described using several crucial features, for example, street name, distance and mobile plan
Slightly (shift strategy is used to indicate that in route segment, at the end of the action to be taken, for example, " steerings ", " continuing traveling ", " from
Open ", " arrival ".This can be calculated based on the relation of the route segment of neighbour.There are these information, one can be generated by turning
The navigation that direction is represented.For example：" 3 kilometers of rear lefts of traveling are rotated into Olympics main road ", 3 kilometers are distances, and right-hand rotation is mobile
Strategy, Olympics main road is street name.
(2) for known route segment, emphasis is how to generate simplicity, it is easy to the description of understanding.For user's warp
The known path often passed through, we describe this section of way by describing its starting terrestrial reference and terminal terrestrial reference, remove centre
Some details, for example, from the home to the route of workspace.If in the case that starting point and destination are not enough to description route, along this
The main roads of bar route running are also included within description.In order that path summarizes more smooth, we define several sentences
Subtemplate describes a known route segment.For example " travel, be designated as by XXX streets until XX from starting point towards destination
Only ... " starting point and destination are starting point and the destination of known paths, and street name is the arterial street name in known paths, ground
Mark represents the terminal of this known paths section.Note, according to feelings actual conditions, it is convenient to omit one or more characteristics, such as starting point
And street name.
Known paths are more, and the route segment produced after being divided to path is fewer, and description gets up also more to save resource, and
Traditional turn direction (turn-by-turn) navigation will not produce change according to this information.
The present invention can be used for giving user one general introduction on route, is generally so short that than the tediously long navigation being made up of turning
It is many.For example, from the home to job site, description can since " being driven to workplace ", this will to user one intuitively build
View goes to determine the driving of next step.When driver is close to the position for deviateing the known route specified, a prompting will be provided, and
Suggestion is provided in the way of turn direction is navigated.
The application scenarios of the present invention：
Application scenarios 1：Known route is used to generate more succinct navigation.Assuming that user plan drive to dining room and by
The corresponding route RB that general navigation system is produced.By analysis of history track data, the present invention it can be found that driver is through frequentation
Ask that supermarket A and corresponding path RA, RA and RB subpath have been overlapped greatly.Using the present invention it is last as a result,
Direct driver instructor is driven to arrive supermarket A, rather than the navigation changed in detail is provided by rotation direction.For the unfamiliar road of driver
Line, still there is provided detailed by turning direction.
Application scenarios 2：Driver comes a new city, can find, be given birth to according to natural route without known route
Navigation into turn direction is supplied to him.
In the present invention, by extracting the historical trajectory data of user, it is considered to which user's goes out row mode so that navigation instruction
More brief introduction, it is more personalized.In this process, the framework summarized is divided and made a summary we have proposed a new path, first
Based on access frequency of the user to path, the optimum division of route is found, then generates a summary to describe each route segment.
Finally, extensive experiment and subjective research really and on the track data collection of synthesis, are being carried out.Test result indicates that,
In most cases, compared with traditional navigation being made up of rotation direction, the framework can be provided the user based on historical track
Navigation, more intuitively, and can be in Mobile solution, the resource needed for substantially reducing be (for example, reducing sentence number reduces navigation
With the resource needed for bandwidth).
Conventional algorithm, input is generally made up of starting point and destination, and output is based on some standards (such as conventional, safety)
Best route.And be a route in the input of the present invention, output is a route summary.Personalization tourism recommendation is one
Individual special navigation system, target is to find a route, best suits the preference of user.Our work is the navigation system of bottom
System, therefore can be integrated with any navigation system.That is, the present invention can take any road produced by navigation system
Line, and be mapped in a pattern travelled based on user's history.
Historical trajectory data is calibrated and clustered first to the framework of the present invention, extracts the knowledge and known road network of user
Network.Next step, for given path, multistage is divided into using the feature in the first step by path.The step is formally to look for
To the route of optimal segmentation, its quality is improved to greatest extent and is laid the foundation.
As the other embodiment of the present invention, as shown in fig. 7, described other embodiment, it is a kind of personalized path
Navigation system, including：Data preprocessing module, path division module and navigation information generation module, data preprocessing module are used
In the historical trajectory data from user, the routing information that user often accesses is extracted；Path division module, it is optimal for finding
Path is divided；Navigation information generation module, the routing information that the user based on extraction often accesses, known to user on path
The personalized path navigation of generation.
It will be appreciated by those of skill in the art that the method step of each example described with reference to the embodiments described herein
Rapid and module, can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually with hard
Part or software mode are performed, depending on the application-specific and design constraint of technical scheme.Professional and technical personnel can be with
Distinct methods are used each specific application to realize described function, but this realization should not exceed the model of the present invention
Enclose.
It is apparent to those skilled in the art that, for convenience of description and succinctly, foregoing description is
The specific work process of system, device and module, may be referred to the corresponding process in preceding method embodiment, will not be repeated here.
Disclosed herein method, system and module, can realize by another way.For example, described above
Device embodiment be only illustrative, it is actual for example, the division of the module, can be only a kind of division of logic function
There can be other dividing mode when realizing, such as multiple module or components can combine or be desirably integrated into another and be
System, or some features can be ignored, or not perform.It is another, shown or discussed coupling or direct-coupling each other
Or communication connection is it may be said that by some interfaces, the INDIRECT COUPLING or communication connection of device or module, can be it is electrical, machinery or
Other forms.
The module that the discrete parts illustrates can be or may not be physically separate, be shown as module
Part can be or can not be physical module, you can with positioned at a place, or can also be distributed to multiple network moulds
On block.Some or all of module therein can be selected according to the actual needs to realize the scheme purpose of the present embodiment.
In addition, each functional module in each embodiment of the invention can be integrated in a processing module, can also
That modules are individually physically present, can also two or more modules be integrated in a module.
If the function is realized using in the form of software function module and is used as independent production marketing or in use, can be with
It is stored in a computer-readable recording medium.Understood based on such, technical scheme is substantially right in other words
The part or the part of the technical scheme that prior art contributes can be embodied in the form of software product, the calculating
Machine software product is stored in a storage medium, including some instructions are to cause a computer equipment (can be personal
Computer, server, or network equipment etc.) perform all or part of step of each of the invention embodiment methods described.And
Foregoing storage medium includes：USB flash disk, mobile hard disk, system memory (Read-Only Memory, ROM), random access memory
Device (Random Access Memory, RAM), magnetic disc or CD etc. are various can be with the medium of store program codes.
Described above is only the preferred embodiment of the present invention, it should be understood that the present invention is not limited to described herein
Form, is not to be taken as the exclusion to other embodiment, and available for various other combinations, modification and environment, and can be at this
In the text contemplated scope, it is modified by the technology or knowledge of above-mentioned teaching or association area.And those skilled in the art are entered
Capable change and change does not depart from the spirit and scope of the present invention, then all should appended claims of the present invention protection domain
It is interior.
Claims (12)
1. the guidance path framework suitable for reducing redundancy navigation, it is characterised in that it includes：Data preprocessing module, path
Module is summarized in division module and path, and data preprocessing module is connected with path division module, and path division module and path are total
Tie module connection；
Described data preprocessing module, for extracting user knowledge；
Described path division module, is divided for finding optimal path；
Module is summarized in described path, for generating personalized path navigation known to user on path；
Described data preprocessing module includes matching and calibration module, terrestrial reference extraction module and user knowledge measurement module,
With being connected respectively with user knowledge measurement module with calibration module, terrestrial reference extraction module, user knowledge measurement module is drawn with path
Sub-module is connected；
Described matching and calibration module, for track calibration and trajectory clustering, extract the routing information that user often accesses, institute
The routing information that the user stated often accesses includes avenue title；
Described terrestrial reference extraction module, for from the historical trajectory data of user, extracting landmark information, described landmark information
Including start node and target point；
Described user knowledge measurement module, the known paths for exporting user.
2. framework according to claim 1, it is characterised in that it includes：Subscriber interface module, described user interface mould
Block is connected with path division module, also, is also connected with path summary module, for inputting path and being summarized for outgoing route
The personalized path navigation of module generation.
3. framework according to claim 1, it is characterised in that it includes：Data source modules, described data source modules with
Data preprocessing module is connected, the pending data for importing；The source of described pending data, including：Track number
Historical trajectory data, interest point data source and the ground applied according to recording equipment, mobile terminal navigation application and vehicle terminal navigation
Chart database.
4. framework according to claim 1, it is characterised in that：Including route known to user and the unfamiliar road of user
In one route of line or a plurality of route, for route known to user, user is instructed using the personalized route guidance；It is right
In the unfamiliar route of user, turn direction navigation instruction user is used.
5. framework according to claim 1, it is characterised in that：Use is weighed by calculating a familiarity fraction f (R)
Family is in the push to route, described familiarity fraction f (R) according to user pass by the path frequency and the path
Terrestrial reference meaning calculate obtain；Wherein, f (R) size interval range includes：f(R)∈[0,1].
6. framework according to claim 5, it is characterised in that：To path segments R (i, j) familiarity, using as follows
Equation calculates familiarity fraction f (R (i, j))：
<mrow>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>(</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>)</mo>
</mrow>
<mo>&CenterDot;</mo>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mfrac>
<mrow>
<mi>l</mi>
<mi>e</mi>
<mi>n</mi>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>(</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>l</mi>
<mi>e</mi>
<mi>n</mi>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
</mrow>
Wherein,It is the length ratio of path segments, g () is a monotonic function, g (0)=0, g (1)=1, R (i,
J) path segments for being R, terminate since R (i) to R (j), and R (i) is path R i-th of terrestrial reference, and R (j) is path R's
J-th of terrestrial reference.
7. framework according to claim 5, it is characterised in that：In a kind of travel navigation scene, to road known to user
Line, using passenger as authority, terrestrial reference carries out the meaning of terrestrial reference using HITS algorithms as center, registration office as hyperlink
Calculate, weighted sum then is normalized in the pass by result of calculation of the frequency in the path and the meaning of terrestrial reference of user, by institute
The result for stating normalization weighted sum is used as the familiarity fraction f (R) of known paths；For natural route, then it is familiar with
Degree fraction f (R) is set to a steady state value.
8. the framework according to claim any one of 1-7, it is characterised in that including step：
S1：From the historical trajectory data of user, path known to extraction user makes user have known path corresponding
It is familiar with value；
S2：Based on road network information architecture natural route；
S3：Optimal path is found using Dynamic Programming to divide, or, first make path proximity using path candidate condition is relaxed
Match somebody with somebody, then find optimal path division；
S4：Path segments after being divided based on path, the personalized path of generation is summarized, and instructs user.
9. framework according to claim 8, it is characterised in that in step sl, including sub-step：
S101：Calibrated using map match with the track based on anchor, initial trace T be converted into a track based on terrestrial reference,
It is used as anchor point by handling terrestrial reference；After calibration, on the basis of track alignedly target, the track of calibration is used as terrestrial reference
Sequence, as path；
S102：After trajectory path has been demarcated, the paired similitude of the trajectory path demarcated is calculated using EDR distance algorithms,
And path cluster is constituted using similar path；
S103：Using equation below, most representative path R in each path cluster is selected*It is used as known paths；
Wherein, R is path,For path cluster, DEDRFor the EDR distances of two paths；
S104：User is weighed using path familiarity fraction f (R) and path segments familiarity fraction f (R (i, j)) to satisfy the need
The familiarity in footpath.
10. framework according to claim 8, it is characterised in that in step s 2, including sub-step：
S201：Input road network, and initialization path clusterMakeFor empty set；
S202：According to the Name ＆ Location of path segments in road network, it is ranked up using comparator, and from road network
According to the Rule Extraction a line of agreement, a line of extraction is assigned to setting variable e1；
S203：The side extracted in step S202 is removed in road network, then, this side of removal is deposited into the R of path；
S204：In remaining road network, searched using path R and belong to same section, be joined directly together and angle of turn
Less than remaining path of setting value, and the link between the path that finds is completed, finally by these paths link composition candidate
Path cluster；
S205：The Name ＆ Location of path segments in path candidate cluster is sorted using comparator, and from step S204
In path candidate cluster according to agreement Rule Extraction a line, a line of extraction is assigned to setting variable
e2；
S206：The side extracted in step S205 is removed from road network；
S207：By path R and find, satisfactory setting variable e2In path connection, constitute new path；
S208：When path R can not find remaining path candidate cluster again, terminate to set variable e1Headed by path link
Process, path cluster is deposited into by the new path foundIn；
S209：Circulation performs step S202~S208, and traversal searches all paths in road network, until not having in road network
Untill having manageable road segment segment, then, return path cluster
11. framework according to claim 8, it is characterised in that in step s3, including：
(1) the use Dynamic Programming described in is found optimal path and divided, including sub-step：
S3012：Whole path R is traveled through, the path candidate cluster that retrieval is linked comprising path has been retrieved after all path candidates,
The condition optimal dividing of the subpath in whole path R is calculated one by one；
S3013：Path is linked to divide to merge with current path and generates new path division, it is new using equation below incremental computations
The quality score that the path of generation is divided：
Wherein, the quality score that Q (i, R'(j, k)) divides for newly-generated path；For path segments R (i-1, i)
Path candidate cluster；(R'(j k) is path segments R'(j, familiarity fraction k) to f；λ is the non-negative that user specifies
Constant；(i-1 i) arrives R (i) path segments to R for R (i-1)；R " (l, m) is the path segments that R " (l) arrives R " (m)；R'(j,k)
For R'(j) arrive R'(k) path segments；R " (l) is path R " l-th of terrestrial reference；R " (m) is path R " m-th of terrestrial reference；R'
(j) j-th of terrestrial reference for being path R'；R'(k) k-th of terrestrial reference for being path R'；
Then, selection quality score highest one is divided as optimal path；
S3014：Circulation performs step S3012~S3014, is constantly iterated calculating, the optimal road until finding whole path R
Footpath is divided, then, is returned to the whole path R optimal paths and is divided.
(2) use described in relaxes path candidate condition and first makees path proximity matching, then finds optimal path division, including sub-step
Suddenly：
S3021：Define route matching relation and define path similarity relation, including：
(a) route matching relation is defined：Two paths R and R` are given, each path is represented by a series of terrestrial reference, a path
Matching M (R, R`) is a terrestrial reference arranged properly to set, and each first terrestrial reference of terrestrial reference centering comes from R, second terrestrial reference
Come from R2；
(b) path similarity relation is defined：Given two paths R and R`, and if only if, and R and R` has common terminal, and exists
One optimal path matching M (R, R`), then R is similar to R`；Described optimal path matching M (R, R`) is met：
Each terrestrial reference in R and R` of I occurs at least one times in M (R, R`)；
Each terrestrial reference of II in M (R, R`) is to (li, li`), their network space DN(li,li`)≤ε；
The terrestrial reference pair intersected is not present in III in M (R, R`)；
Wherein, ε is the maximum road network spacing between the terrestrial reference pair of matching；
S3022：It is that last path segments is R'(j to set Q (i, R'(j, k)), R (1, the optimal path relaxed i) k)
The quality score of division, when R (i) is a R terminal, regulation R (i)=R` (k), then the optimal path relaxed, which is divided, to be met
Following recurrence formula：
The condition optimal path relaxed is calculated to divideIncluding following two situations：
If 1. k-j>1, then the condition optimal path relaxed is divided intoWithIt is last
One path segments comes from same paths, enumerates as R (i-1) matching R` (k), R (i-1) matching R` (k-1), R (i) matchings R`
(k-1) three kinds of situations when, then incrementally calculate the quality score Q (i, R'(j, k)) of path division using equation below：
Q(i,R'(j,k))←max{Q(i-1,R'(j,k)),Q(i-1,R'(j,k-1))+f(R'(j,k))-f(R'(j,k-
1)),
Q(i,R'(j,k-1))+f(R'(j,k))-f(R'(j,k-1))}
If 2. k-j=1, that is, terminate the search procedure of path candidate, a new route segment R` (j, k), time now are opened
Routing footpath clusterR` (j) is contained, also, R`` isIn arbitrary path, for a path segments R``
, there are two kinds of situations in (l, m), wherein R`` (m)=R` (j)：R`` (m) and R (i-1) is matched or R`` (m) and R (i) is matched, piece
Both of these case is lifted, the quality score Q (i, R'(j, k)) of path division is then incrementally calculated using equation below：
Then, selection quality score highest one is divided as optimal path.
12. framework according to claim 8, it is characterised in that in step s 4, including：
(1) for natural route section, it is described using multiple key features；Described multiple key features include street name,
Distance and shift strategy, described shift strategy represent the action taken at the end of natural route section, and it includes turning to, continued
Travel, leave and reach；
(2) for known paths section, it is described by describing its starting terrestrial reference and terminal terrestrial reference, in starting point terrestrial reference and terminal
Ground terrestrial reference is not enough in the case of describing it, and the turnpike road travelled along this paths is also included within description；According to reality
Situation, can omit one or more features, and the feature of omission includes starting point, street name.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710283551.7A CN107084735A (en) | 2017-04-26 | 2017-04-26 | Guidance path framework suitable for reducing redundancy navigation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710283551.7A CN107084735A (en) | 2017-04-26 | 2017-04-26 | Guidance path framework suitable for reducing redundancy navigation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107084735A true CN107084735A (en) | 2017-08-22 |
Family
ID=59612183
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710283551.7A Pending CN107084735A (en) | 2017-04-26 | 2017-04-26 | Guidance path framework suitable for reducing redundancy navigation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107084735A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107544516A (en) * | 2017-10-11 | 2018-01-05 | 苏州大学 | Automated driving system and method based on relative entropy depth against intensified learning |
| CN109255085A (en) * | 2018-04-28 | 2019-01-22 | 云天弈（北京）信息技术有限公司 | A kind of search result shows system and method |
| WO2019051834A1 (en) * | 2017-09-18 | 2019-03-21 | Baidu.Com Times Technology (Beijing) Co., Ltd. | Driving scenario based lane guidelines for path planning of autonomous driving vehicles |
| CN109556614A (en) * | 2018-12-14 | 2019-04-02 | 北京百度网讯科技有限公司 | Localization method and device for automatic driving car |
| CN110674723A (en) * | 2019-09-19 | 2020-01-10 | 北京三快在线科技有限公司 | Method and device for determining driving track of unmanned vehicle |
| CN110968659A (en) * | 2019-12-05 | 2020-04-07 | 湖北工业大学 | Redundancy removing method for high-rise navigation road network based on continuous road chain |
| CN113218412A (en) * | 2021-04-29 | 2021-08-06 | 重庆长安汽车股份有限公司 | Method, device and storage medium for familiar destination exit navigation |
| CN113988482A (en) * | 2021-12-23 | 2022-01-28 | 深圳市领志光机电自动化系统有限公司 | Three-dimensional space track path automatic matching generation system based on internet data chain |
-
2017
- 2017-04-26 CN CN201710283551.7A patent/CN107084735A/en active Pending
Non-Patent Citations (2)
| Title |
|---|
| YAGUANG LI 等: "PerNav: A Route Summarization Framework for Personalized Navigation", 《SIGMOD "16 PROCEEDINGS OF THE 2016 INTERNATIONAL CONFERENCE ON MANAGEMENT OF DATA》 * |
| YAGUANG LI等: "PaRE: A System for Personalized Route Guidance", 《 WWW "17 PROCEEDINGS OF THE 26TH INTERNATIONAL CONFERENCE ON WORLD WIDE WEB》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019051834A1 (en) * | 2017-09-18 | 2019-03-21 | Baidu.Com Times Technology (Beijing) Co., Ltd. | Driving scenario based lane guidelines for path planning of autonomous driving vehicles |
| CN110325935A (en) * | 2017-09-18 | 2019-10-11 | 百度时代网络技术（北京）有限公司 | The lane guide line based on Driving Scene of path planning for automatic driving vehicle |
| US10807599B2 (en) | 2017-09-18 | 2020-10-20 | Baidu Usa Llc | Driving scenario based lane guidelines for path planning of autonomous driving vehicles |
| CN110325935B (en) * | 2017-09-18 | 2022-08-12 | 百度时代网络技术（北京）有限公司 | Driving scenario based lane guide line for path planning of autonomous vehicles |
| CN107544516A (en) * | 2017-10-11 | 2018-01-05 | 苏州大学 | Automated driving system and method based on relative entropy depth against intensified learning |
| CN109255085A (en) * | 2018-04-28 | 2019-01-22 | 云天弈（北京）信息技术有限公司 | A kind of search result shows system and method |
| CN109556614A (en) * | 2018-12-14 | 2019-04-02 | 北京百度网讯科技有限公司 | Localization method and device for automatic driving car |
| CN110674723A (en) * | 2019-09-19 | 2020-01-10 | 北京三快在线科技有限公司 | Method and device for determining driving track of unmanned vehicle |
| CN110968659A (en) * | 2019-12-05 | 2020-04-07 | 湖北工业大学 | Redundancy removing method for high-rise navigation road network based on continuous road chain |
| CN113218412A (en) * | 2021-04-29 | 2021-08-06 | 重庆长安汽车股份有限公司 | Method, device and storage medium for familiar destination exit navigation |
| CN113988482A (en) * | 2021-12-23 | 2022-01-28 | 深圳市领志光机电自动化系统有限公司 | Three-dimensional space track path automatic matching generation system based on internet data chain |
| CN113988482B (en) * | 2021-12-23 | 2022-03-15 | 深圳市领志光机电自动化系统有限公司 | Three-dimensional space track path automatic matching generation system based on internet data chain |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106931986A (en) | Personalized method for path navigation and system | |
| CN107084735A (en) | Guidance path framework suitable for reducing redundancy navigation | |
| CN109977283A (en) | A kind of the tourism recommended method and system of knowledge based map and user's footprint | |
| CN105069047B (en) | A kind of search method and device of geography information | |
| Moncla et al. | Reconstruction of itineraries from annotated text with an informed spanning tree algorithm | |
| CN107423837A (en) | The Intelligent planning method and system of tourism route | |
| CN107228677A (en) | Driftage recognition methods and device | |
| CN107391706A (en) | A kind of city tour's question answering system based on mobile Internet | |
| Li et al. | Pare: A system for personalized route guidance | |
| CN109992786A (en) | A kind of semantic sensitive RDF knowledge mapping approximate enquiring method | |
| CN106909611A (en) | A kind of hotel's automatic matching method based on Text Information Extraction | |
| Sacharidis et al. | Routing directions: Keeping it fast and simple | |
| Chia et al. | Trip planning route optimization with operating hour and duration of stay constraints | |
| Wang et al. | Personalized route recommendation with neural network enhanced search algorithm | |
| Zhao et al. | A deep inverse reinforcement learning approach to route choice modeling with context-dependent rewards | |
| Mokhtari et al. | Tagging address queries in maps search | |
| CN113392147B (en) | VR scene knowledge graph representation and dynamic update method | |
| CN106155998A (en) | A kind of data processing method and device | |
| CN114462357A (en) | Entity linking method, entity linking device, entity constructing method and entity linking device and storage medium | |
| CN110532464A (en) | A kind of tourism recommended method based on more tourism context modelings | |
| Liao et al. | Fusing geographic information into latent factor model for pick-up region recommendation | |
| Li et al. | PerNav: A route summarization framework for personalized navigation | |
| Zhou et al. | Smart tour route planning algorithm based on clustering center motive iteration search | |
| Su et al. | Personalized route description based on historical trajectories | |
| Chen et al. | Modeling tourism using spatial analysis based on social media big data: A review |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170822 |