CN108961758A - A kind of crossing broadening lane detection method promoting decision tree based on gradient - Google Patents
A kind of crossing broadening lane detection method promoting decision tree based on gradient Download PDFInfo
- Publication number
- CN108961758A CN108961758A CN201810877633.9A CN201810877633A CN108961758A CN 108961758 A CN108961758 A CN 108961758A CN 201810877633 A CN201810877633 A CN 201810877633A CN 108961758 A CN108961758 A CN 108961758A
- Authority
- CN
- China
- Prior art keywords
- lane
- road
- section
- floating car
- track
- 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.)
- Granted
Links
- 238000003066 decision tree Methods 0.000 title claims abstract description 37
- 238000001514 detection method Methods 0.000 title claims abstract description 26
- 230000001737 promoting effect Effects 0.000 title claims abstract description 10
- 238000007667 floating Methods 0.000 claims abstract description 114
- 238000000034 method Methods 0.000 claims abstract description 41
- 238000004458 analytical method Methods 0.000 claims abstract description 19
- 238000011160 research Methods 0.000 claims abstract description 18
- 230000008859 change Effects 0.000 claims abstract description 14
- 239000006185 dispersion Substances 0.000 claims abstract description 9
- 238000005192 partition Methods 0.000 claims abstract description 7
- 238000000926 separation method Methods 0.000 claims abstract description 5
- 230000011218 segmentation Effects 0.000 claims description 25
- 238000012549 training Methods 0.000 claims description 10
- 239000000523 sample Substances 0.000 description 15
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000002474 experimental method Methods 0.000 description 5
- 239000000284 extract Substances 0.000 description 5
- 230000033001 locomotion Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000007637 random forest analysis Methods 0.000 description 3
- 238000007619 statistical method Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 241000208340 Araliaceae Species 0.000 description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010835 comparative analysis Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 235000008434 ginseng Nutrition 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- NPYPAHLBTDXSSS-UHFFFAOYSA-N Potassium ion Chemical compound [K+] NPYPAHLBTDXSSS-UHFFFAOYSA-N 0.000 description 1
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 229910052700 potassium Inorganic materials 0.000 description 1
- 229910001414 potassium ion Inorganic materials 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000007790 scraping Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0108—Measuring and analyzing of parameters relative to traffic conditions based on the source of data
- G08G1/0112—Measuring and analyzing of parameters relative to traffic conditions based on the source of data from the vehicle, e.g. floating car data [FCD]
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
Landscapes
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention provides a kind of crossing broadening lane detection method that decision tree is promoted based on gradient, including carrying out data input, signature analysis and feature selecting, the dispersion of distribution, distribution density, the angular separation of FCD and speed of the floating car data point on road cross section are selected；Basic classification device is constructed, including the Run-time scenario according to floating wheel paths, constructs basic classification device；The lane for promoting decision tree based on gradient calculates, including the use of floating car data, referring to basic classification device, Decision-Tree Method is promoted according to gradient, when FCD is divided different lane quantity by selection, comentropy is minimum and the driveway partition is optimally distributed the corresponding classification number of state as the corresponding number of lanes of this section of road on cross section；It broadens lane to determine, including on the same road of comparison, lane quantity situation of change on the corresponding road section of difference research unit judges crossing with the presence or absence of broadening lane.The present invention improves the precision of road junction roadway number judgement.
Description
Technical field
The invention belongs to space-time trajectory big data technical fields, are related to a kind of crossing broadening that decision tree is promoted based on gradient
Lane detection method.
Background technique
With popularizing for GPS positioning technology, more and more vehicles are mounted with GPS receiver, and the track of this kind of vehicle is several
Cover whole road networks, road surface [1,2] in city, these space-time GPS track data, for the road letter based on floating car data
Breath acquiring technology provides the historical data of magnanimity.
City road network is extracted using space-time GPS track big data, road track information is the hot topic studied at present.This
Save summarized from data source in terms of road track quantity based on floating wheel paths calculate and intersection broaden lane the two
The Developments of aspect summarize research achievement at this stage, and the problem of exist at this stage.
1) the road track quantity based on floating wheel paths calculates
Since Floating Car always travels on road, practical driving trace sketches the contours of the topological structure and vehicle of city road network
Road modification information etc., more and more researcher start with floating wheel paths and obtain urban road information both at home and abroad.Due to
The GPS positioning device precision of different vehicle installation is different, therefore, existing to be based on FCD (Floating Car Data, Floating Car
Data) the method that calculates of road track quantity can be divided into according to the positioning accuracy of vehicle itself: based on high quality GPS number
According to road track quantity calculate and road track quantity based on ubiquitous FCD data calculates.
(1) the road track quantity based on high quality GPS data calculates
Road track quantity computation method based on high quality GPS data, data source are usually differential GPS
High-precisions GPS track data such as (Difference Global Positioning System, DGPS).Wagstaff K. etc.
The method for extracting road axis and lane line using space-time DGPS track data is explored, and is calculated using K mean value
Method carries out clustering processing to the track DGPS, obtains lane information [3] from cluster result；Thereafter, Fang etc. proposes a kind of de-
Method for obtaining road information from original map is utilizing height by taking track classification-cluster method to obtain road information
This mixed model (Gaussian mixture model) obtains road track information [4]；Chen Y etc. using track classification and
Fusion method extracts road grade information from a large amount of DGPS track datas, extracts lane information [5] using gauss hybrid models；
Edelkamp etc. uses hierarchical clustering algorithm and is first clustered to DGPD track data and classifies again, using the cluster center of cluster as
Lane center line position, and using the classification number of cluster as number of lanes [6]；Knoop V L etc. utilizes outfit GPS-PPP
The track of vehicle data of (Static Precise Point Positioning) estimate position and the width [7] in different lanes.The studies above utilizes high quality
GPS track data acquisition road information extracts lane information, complex cross crossing geometry etc. that road is arrived in fining, but
The problems such as such method is required to professional equipment, and there are data acquisition cost height, update cycle length, complex procedures.
(2) the road track quantity based on ubiquitous FCD data calculates
Currently contain Traffic Information abundant in the Floating Car track data of ubiquitous, magnanimity, low cost,
Uduwaragoda etc. utilizes cuclear density clustering method, and lane quantity and lane position [8] are detected from GPS track data.Wang
City lane rank road network information is extracted Deng the floating car data using high sample frequency, and emphasis has been explained and floated from low precision
The extracting method [9] of intersection complexity road network is detected in car data.Tang's furnace is bright equal to FCD progress signature analysis, has studied base
The lane quantity acquisition methods [1-2] of mixed model are constrained in Naive Bayes Classification method and Gauss.The method achieve from
Quick obtaining lane quantity information in floating car data, but its section lane quantity misjudgment rate near road junction
Higher, reason is not fully consider intersection traffic operation characteristic and its unique geometry.Above-mentioned utilization is ubiquitous
The research that the road track quantity of FCD calculates, is all based on space-time trajectory category theory.
In the track sort research based on GPS, Jahangiri etc. [10] and Shafique etc. [11] using decision tree,
Support vector machines, naive Bayesian, conditional random fields, K be neighbouring, Bagging, random forest and GBDT (Gradient
Boost Decision Tree, gradient promoted decision tree) etc. taxonomic methods division experiment is carried out to track, achieve preferably
Classifying quality, wherein Shafique etc. [11] has found random forest and GBDT nicety of grading highest.However, these tracks point
The research of class mainly classified using GPS data to trip mode, not by based on integrated study random forest and
GBDT classification method is applied in the division of road track information.
2) research of intersection broadening road traffic
With the fast development in city, the pressure that road traffic system is faced is gradually increased, logical to improve intersection
Row ability, broadening lane become a kind of excessive measure for reducing signal cross crossing congestion degree.According to statistics, in existing crossroad
In mouthful, it is generally provided with broadening lane [12].Foreign countries are concentrated mainly on choosing lane model for the research in broadening lane and lead to
In terms of the analysis of row ability [13-14], Moon J P etc. studies the operation evaluation problem for widening lane, and to merging area
Traffic safety assessed [12].Bright wait of Tang's furnace utilizes low frequency floating car data, automatic identification city road network intersection
And extract the intersection detailed structure [15] under road network.
Current research primarily focuses on broadening section to be influenced by intersection capacity, and data are all from artificial tune
The method looked into.Since its data volume is limited, so that the truth of intersection can not be reacted completely.At the same time, domestic
There is relevant regulation to intersection Widening Design in existing specification, but existential specification standard is different, there is conflict each other
The problem of contradiction [1].Existing research the result shows that, although broadening lane can improve the passage of intersection to a certain extent
Ability, but due to its special geometrical condition and operation characteristic, it can often cause the road traffic accidents [12,16] such as scraping.Due to
The magnanimity and real-time of floating car data, can be to intersection exhibition using Floating Car track data combined data digging technology
Wide road detects in time, and to analyze crossing situation of change, auxiliary road network navigation data updates.
Bibliography
[1] Tang's furnace is bright, Yang Xue, Kan Zihan, waits a kind of lane population detection [J] China based on Naive Bayes Classification of
Highway journal, 2016,29 (3): 116-123.
[2] Tang's furnace is bright, Yang Xue, Jin Chen, and is waited to obtain [J] Wuhan University based on the lane information of constraint gauss hybrid models
Journal information science version, 2017,42 (3): 341-347.
[3]Wagstaff K,Cardie C,Rogers S,et al.Constrained k-means clustering
with background knowledge[C]//ICML.2001,1:577-584.
[4]Fang L N,Yang B S.Automated extracting structural roads from
mobile laser scanning point clouds[J].Acta Geodaetica et Cartographica
Sinica,2013,42(2):261-267.
[5]Chen Y,Krumm J.Probabilistic modeling of traffic lanes from GPS
traces[C]//Proceedings of the 18thSIGSPATIAL International Conference on
Advances in Geographic Information Systems.ACM,2010:81-88.
[6]Edelkamp S,S.Route planning and map inference with global
positioning traces[J].Computer Science in Perspective,2003:128-151.
[7]Knoop V L,de Bakker P F,Tiberius C C J M,et al.Lane Determination
With GPS Precise Point Positioning[J].IEEE Transactions on Intelligent
Transportation Systems,2017.
[8]Uduwaragoda E R I A C,Perera A S,Dias S A D.Generating lane level
road data from vehicle trajectories using Kernel Density Estimation[C]//
International IEEE Conference on Intelligent Transportation Systems.IEEE,
2014:384-391.
[9]Wang J,Rui X,Song X,et al.A novel approach for generating routable
road maps from vehicle GPS traces[J].International Journal of Geographical
Information Systems,2015,29(1):69-91.
[10]Jahangiri A,Rakha H A.Applying Machine Learning Techniques to
Transportation Mode Recognition Using Mobile Phone Sensor Data[J].IEEE
Transactions on Intelligent Transportation Systems,2015,16(5):2406-2417.
[11]Shafique M A,Hato E.A Comparison among various Classification
Algorithms for Travel Mode Detection using Sensors'data collected by
Smartphones[C]//International Conference on Computers in Urban Planning and
Urban Management.2015.
[12]Moon J P,Reese P E,Michael P,et al.Evaluation of operations and
safety in a congested freeway merging area with auxiliary through lane[R]
.2012.
[13]Tarawneh M S,Tarawneh T M.Effect on utilization of auxiliary
through lanes of downstream right-turn volume[J].Journal of transportation
engineering,2002,128(5):458-464.
[14]Moriyama Y,Mitsuhashi M,Hirai S,et al.The effect on lane
utilization and traffic capacity of adding an auxiliary lane[J].Procedia-
Social and Behavioral Sciences,2011,16:37-47.
[15] Tang's furnace is bright, Niu Le, Yang Xue, and is waited to carry out urban road junction identification and structure using track big data
It extracts [J] and surveys and draws journal, 2017,46 (6): 770-779.
[16] horse is gorgeous, Gao Yuee, cold snow, and signal cross crossing is waited to broaden the Harbin lane traffic operation characteristic [J]
Polytechnical university's journal, 2015,47 (2): 42-45.
[17] Wen Changbao, Koryo is red, Fang Jishan, waits based on the high-precision weighing system of modified clipping average filter method
Study [J] sensing technology journal, 2014 (5): 649-653.
[18] GB50647-2011, urban road junction planning Plan Press of specification China, Beijing, 2011.
[19] Wuhan research [D] of behavior formula identifying code man-machine identification of the Su Tao based on gradient boosted tree: Central China is pedagogical
University, 2016.
[20]Friedman J H.Greedy function approximation:a gradient boosting
machine[J].Annals of statistics,2001:1189-1232.
[21] practice and understanding of Liu Zhang temperature akaike information criterion AIC and its meaning [J] mathematics, 1980 (3): 65-
73.
Summary of the invention
It is an object of the invention to overcome prior art defect, a kind of crossing broadening that decision tree is promoted based on gradient is provided
Lane detection method.
Technical solution of the present invention provide it is a kind of based on gradient promoted decision tree crossing broaden lane detection method, including with
Lower step:
Step 1, data input, including input floating car data and urban road polar plot carry out data prediction, reject
Then shift point in floating wheel paths divides floating car data into floating at the floating car data and non-crossing mouth of intersection
Motor-car data, then the road covered respectively according to track are equidistantly divided along road direction, obtain several compartmenteds as base
This research unit；
Step 2, signature analysis, including intersection lane broaden Variations, and with intersection lane variation pair
The FCD track characteristic analysis answered；
Step 3, feature selecting, including selecting 4 characteristic parameters for constructing basic classification device and the calculating of number of track-lines amount,
The respectively dispersion of distribution, distribution density, the angular separation of FCD and speed of the floating car data point on road cross section；
Step 4, basic classification device is constructed, including the Run-time scenario according to floating wheel paths, building is based on FCD probability density
Lane quantity basic classification device；
Step 5, the lane for promoting decision tree based on gradient calculates, including the use of floating car data, referring to basic classification device,
Decision-Tree Method is promoted according to gradient, when FCD is divided different lane quantity by selection, comentropy minimum and the driveway partition
The corresponding classification number of the state that is optimally distributed on cross section is as the corresponding number of lanes of this section of road；
Step 6, broadening lane determines, including on the same road of comparison, and the corresponding road section of difference research unit is got on the bus
Road quantity situation of change judges crossing with the presence or absence of broadening lane.
Moreover, in step 4, using the floating car data covered on the section of known lane quantity as training sample, root
According to descruotuve statu statistical method, the flat of the 4 characteristic parameters difference of floating car data covered on the section of different lane quantity is counted
Mean value realizes building basic classification device.
Moreover, the dispersion of distribution of the floating car data point on road cross section, is by calculating each floating in segmentation section
The Euclidean distance value of car data point to track centers obtains.
Moreover, distribution density extracting mode of the floating car data point on road cross section is as follows,
The trace centerline horizontal direction along segmentation section, by floating car data coverage area by being equidistantly divided into several areas
Between, it calculates points amount in track in each section and divides the ratio between total quantity in section, obtain the track dot density in each section；It will be any
The track dot density in section where the corresponding tracing point of floating car data point, as the floating car data point on road cross section
The value of distribution density.
The present invention is that lane grade navigation data updates it is considered that the crossing broadening lane information that quick detection changes often
One of groundwork；However, using the probe vehicles such as profession measurement, image or video interpretation, the analysis of high quality GPS track data
The method of road quantity, it is long there are the period, it is costly the problems such as, from the feature of the floating car data spatial distribution of intersection, with
And the actual demand of road track quantity detection is set out, and in existing road lane on the basis of quantity detection method, is handed over from road
The visual angle of cross road mouth fully considers the mapping relations of the geometry and FCD in intersection lane in intersection lane, this hair
The bright main method for promoting decision tree using gradient, has inquired into the road junction roadway change detection method based on FCD, has improved crossing
The precision of number of track-lines judgement.
Detailed description of the invention
Fig. 1 is the flow chart of the embodiment of the present invention.
Specific embodiment
In order to make the objectives, technical solutions and advantages of the present invention clearer, with reference to the accompanying drawings and examples, to this
Inventive technique scheme is further described.
Referring to Fig. 1, the embodiment of the present invention provides a kind of crossing broadening lane detection method that decision tree is promoted based on gradient:
Step 1: data input, data source includes using urban taxi as the floating car data of carrier and urban road vector
Figure.
Data collection can be carried out when specific implementation in advance, since taxi traveling is in urban road, track is almost
The road network road surface in the city covered, and hiring out car data is to be managed collectively by taxi company, therefore selection is hired out with city
Vehicle is the floating car data of carrier.GPS location, i.e. longitude and latitude or (x, y) coordinate can be used in Floating Car track data.
After input data, in order to improve data precision, it can choose and further original floating car data is pre-processed,
And existing modified clipping average filter algorithm [17] is used, reject the shift point of original floating wheel paths.It is first in embodiment
First to using urban taxi as the floating car data of carrier and urban road polar plot, data prediction is carried out, wherein Floating Car
The elimination of rough difference of data uses existing modified clipping average filter algorithm, rejects the shift point in floating wheel paths, then
According to survey region type, floating car data is divided into 2 classes: the Floating Car at the floating car data and non-crossing mouth of intersection
Data compare and analyze research, then by different types of floating car data, according to the road that its track is covered, along road
Direction is equidistantly divided with 10m, obtains the basic research unit of experimental study, i.e. segmentation section.
Step 2: signature analysis, including intersection lane broaden Variations, and with intersection lane variation pair
The FCD track characteristic analysis answered.
The embodiment of the present invention proposes, analyzes first intersection lane broadening variation characteristic, then analysis is floated
Car data at the intersection, when lane variation/lane does not change Floating Car show feature, finally with the floating at non-crossing mouth
The performance characteristic of wheel paths data compares and analyzes, and the variation for obtaining floating wheel paths at the intersection/non-crossing mouth is special
Sign rule.
When it is implemented, can analyze intersection lane broadening variation characteristic, since road junction adds broadening
Lane can generate lane quantity compared to the road middle section road for belonging to road and change；Then, using cluster --- sorting algorithm,
Analyze the track characteristic variation that lane broadening changes Floating Car in corresponding road surface, including the variation of track geometrical characteristic and rail
The variation of mark motion feature.The method combined using descriptive statistic with document analysis is summarized the broadening of intersection lane and floated downward
The changing rule of motor-car track.Finally, according to Floating Car trajectory coordinates position, to the attributive character of each floating car data point,
The variation tendency of corresponding road different sections of highway compares and analyzes.For step 3, step 4 based theoretical.
Step 3: feature selecting, including selecting several features ginseng for constructing basic classification device and the calculating of number of track-lines amount
Number.
First from the practical problem using floating car data detection intersection expanded lane, according to the feature of previous step
Analysis chooses 4 parameters that gradient promotes decision tree modeling as a result, in conjunction with track motion feature and track geometrical characteristic.
In embodiment, first from the practical problem using floating car data detection intersection expanded lane (that is, being directed to
The invention solves two main problems: using floating car data calculate lane quantity and using floating car data detect crossing
Broaden lane)；Then, according to the signature analysis of previous step as a result, being gradient in conjunction with track motion feature and track geometrical characteristic
Promote 4 parameters of selection of decision tree modeling: dispersion of distribution d of the floating car data point on road cross section, distribution density ρ,
The angular separation ɑ and speed v of FCD,
If dividing includes n FCD tracing point: { (x in section X1,y1),(x2,y2),…,(xn,yn), xi={ di,ρi,αi,
viIt is i-th of track point feature, i=1,2 ..., n.Obtain the characteristic parameter collection { x of basic research unit1,x2,…,xn,
In, xi={ di,ρi,αi,vi}.The feature average value of Floating Car tracing point in research unit is calculated, that is, chooses FCD in segmentation section
It is evenly distributed width, be evenly distributed density, mean direction angle and average speed, be denoted asWherein To wheel paths of floating in descriptive study unit
The overall condition of feature constructs basic classification device for step 4 and establishes data basis.
The present invention solves its Euclidean distance for arriving track fitting center line using will divide the track data in section, and combines
Existing basis road network information, carries out width detection to the floating car data covered in target road section, while recording different in width
In the case of, the density of tracing point meets the tracing point under the conditions of current width and accounts for ratio total in segmentation section.
Step 4: building basic classification device including the Run-time scenario according to floating wheel paths, and combines the road of the existing country
Related codes and standards is built on road, constructs the lane quantity basic classification device based on FCD probability density, that is, is based on floating car data
Test block lane population detection basic classification device.
The basic classification device constructed in embodiment is that the floating car data covered on the section using known lane quantity is made
The floating car data feature ginseng covered on the section of different lane quantity is counted according to descruotuve statu statistical method for training sample
Several average value constructs the test block lane population detection basic classification device based on floating car data.
It due to being modeled using GBDT algorithm, needs that classification number is provided previously, it is therefore desirable to utilize known number of track-lines
The floating car data of type is measured as training sample, floating data (characteristic parameter) is constructed using statistical method, with place road surface
The one-to-one basic classification device of lane quantity, modeled for GBDT, provide may ownership classification number (e.g., referring to basic
Classifier, thus it is speculated that may be 3 lanes or 4 lanes).
FCD is covered on the lane information [1] that the potassium ion distribution in road surface characterizes the road to a certain extent, according to floating
Upper situation of the motor-car data cover on road surface, the lane quantity that may have for speculating the road surface.Due to intersection road
The geometry and traffic circulation characteristic of mouth are different from road middle section, therefore in order to keep sample data representative, the present invention
According to the Run-time scenario of floating wheel paths, and the related codes and standards of road construction of the existing country is combined, FCD probability will be based on
The lane quantity basic classification device of density is divided into two classes: intersection section, non-crossing crossing section.Thereafter, to every class
The average value of training sample extraction featureDetection result and quantity building in lane known to sample is close based on FCD probability
The lane classifier of degree.So that sample to be tested (the segmentation section of unknown lane quantity) is according to the FCD space characteristics collection for dividing section
Average valueReferring to basic classification device, thus it is speculated that the lane quantity that may belong to, to substitute into GBDT model validation lane
Quantity.
Step 5: lane quantity based on GBDT obtains, and calculates lane quantity using floating car data, using GBDT model,
Decision-Tree Method is promoted according to the gradient that document [19,20] propose, essentially by floating car data on analysis road surface
Distribution, into the classification number to speculate GBDT model, with the GBDT model after the combination of this comparison different classifications, choosing will
When FCD divides different lane quantity, the corresponding classification of state that comentropy is minimum and the driveway partition is optimally distributed on cross section
Number is as the corresponding number of lanes of this section of road.
Embodiment is pushed away according to the characteristic parameter average value of floating car data in research unit referring to basic classification device first
The lane quantity (such as 3 lanes or 4 lanes) that the road surface of floating car data covering may have is surveyed, is pushed away referring next to basic classification device
The lane quantity (i.e., thus it is speculated that the classification number of GBDT model) of survey, using GBDT model, the ladder for selecting document [19,20] to propose
Degree promotes Decision-Tree Method, the GBDT model after constructing different classifications combination respectively, finally, comparative analysis is by GBDT points
After class, the distribution on each lane of floating car data on road surface, with the GBDT mould after the combination of this comparison different classifications
Type, when FCD is divided different lane quantity by selection, comentropy is minimum, and the driveway partition is corresponding in basic research unit
On road cross section, the corresponding classification number of optimal distribution state obtains road to be measured as the corresponding number of lanes of this section of road
The lane quantity of section.
In embodiment, it is as follows that implementation is calculated based on the lane that gradient promotes decision tree:
GBDT (Gradient Boost Decision Tree) [19,20] is one nonlinear model, is integrated
One of representative method of habit.Algorithm idea of the GBDT based on boosting in integrated study scope, each iteration are all residual in reduction
The gradient direction of difference creates a decision tree, and the accuracy of classification is continuously improved by iteration.The present invention is it is considered that in practical feelings
In condition, due to the error of urban taxi positioning and the influence of vehicle movement, there is certain drift in the trajectory location points of acquisition
It moves, distribution of the tracing point on road surface is not directly to reflect specific lane quantity, but each tracing point is fallen on lane
Degree of closeness defer to geometry distribution, using track distribution establish rational model to obtain lane quantity.
The track data cover width obtained by detection, in conjunction with the test block lane population detection based on FCD probability density
Basic classification device, using GBDT model, to confirm final lane quantity.Lane quantity based on FCD, which obtains, uses GBDT model,
According to gradient promoted Decision-Tree Method [20], essentially by analysis road surface on floating car data distribution, into
The classification number of GBDT model is speculated, with the GBDT model superiority and inferiority after the combination of this comparison different classifications, when Selection Model is optimal pair
The classification number answered is as the corresponding number of lanes of this section of road.If dividing includes n FCD tracing point: { (x in section X1,y1),
(x2,y2),…,(xn,yn), it is known that there is K lane, x in the road surface of this n point coveringi={ di,ρi,αi,viIt is i-th
A track point feature, i=1,2 ..., n.
Firstly, being the dispersion of distribution of floating car data by the space coordinate conversion of Floating Car, that is, calculate each in segmentation section
Floating car data point to track centers Euclidean distance value, so that coordinate position is converted to Floating Car dispersion of distribution di。
Then, the trace centerline horizontal direction along segmentation section, floating car data coverage area is equidistantly divided by spacing d'
100 sections are segmented into, points amount in track in each section is calculated and divide the ratio between total quantity n in section, obtain the tracing point in each section
Density, using the track dot density value in section where i-th of tracing point as ρiValue, pitch d'=Dmax/ 100,
DmaxTo divide the maximum width that floating wheel paths cover in section, (in some segmentation section, dupMore than trace centerline, distance
The distance of the farthest FCD point of center line；ddownFor trace centerline hereinafter, the distance of the farthest FCD point of distance center line；FCD
Cover width D on road cross sectionmax=| dup|+|ddown|)。
Finally, seeking αiAnd vi, αiIt is the headstock deflection of i-th of FCD point, viIt is the speed of i-th of FCD point.
yiFor the carriageway type (e.g., the 1st lane) where i-th point, loss function is set to L [yi, F (xi)], F (xi)
It is denoted as the carriageway type that tree-model [20] identification is promoted by gradient, the number of iterations is M (i.e. M tree), can be divided into K class (that is, should
Segmentation section has K lane), then
0th time iteration initialization gradient promotes tree-model F(0)(xi) it is definite value, i.e.,
In formula: xi={ di,ρi,αi,viIt is i-th of track point feature, γ is the carriageway type of current iteration initialization；Often
Secondary iteration finds decision tree, and the loss of sample to be allowed to become smaller as far as possible, the m times iteration, i.e. the m decision tree h(m-1)(xi) draw
The best residual error match value for dividing result is F(m)(xi), the match value of the m-1 decision tree division result is F(m-1)(xi), residual error
γikFor
New training set { (x is obtained by formula (2)i,γik) (i=1,2 ..n), as the m+1 decision tree h of training(m)
(xi) when, corresponding leaf node region is Rik, k=1,2 ..K, k calculate area foliage most for identifying kth lane
Good match value
Wherein, xi∈Rik, c indicates the step-length along gradient descent direction, by decision tree h(m)(xi) and formula (3) leaf area
The best-fit values in domainObtain the m times more new model are as follows:
Final decision tree-model F is obtained by maximum number of iterations M times(M)(xi).Choose the work of known lane quantity FCD
For training sample, under the conditions of obtaining different classifications number (i.e. lane quantity K), corresponding gradient promotes decision-tree model.To obtain
The lane quantity for taking lane segmentation section to be measured, according to the FCD feature average value of the segmentation sectionReferring to basic classification
Device, thus it is speculated that the candidate value of the number of lanes of section ownership (such as: the segmentation section may have 2 lanes or 3 lanes).It utilizes
The GBDT model of different classifications combination, the FCD divided in section is divided.Due to the FCD in each segmentation section, correspondence is multiple
GBDT model partition result.It is logical that various combination category of model result is counted, calculate floating car data in the segmentation section
Comentropy:
Num=arg min F (k) (6)
Wherein, AIC is akaike information criterion, it is built upon on the basis of entropy, can be used to measure estimation model
The Optimality [21] of complexity and data fitting；Num is when FCD is divided different lane quantity, most preferably to divide on cross section
Lane quantity corresponding to cloth state is the number of lanes of the segmentation section；PiIt is promoted in decision-tree model for gradient, i-th of track
Point belongs to the probability in kth lane, is detailed in bibliography [1], i.e. the expression formula calculation formula of the Probability p k (x) of kth class.
It is optimally distributed on cross section when FCD is divided different lane quantity by selection in conjunction with statistical result and comentropy
Lane quantity corresponding to state is the number of lanes (Num) of the segmentation section.Step 6: broadening lane determines, according to road junction roadway
Quantity variation judges crossing with the presence or absence of broadening lane.
Divided in conjunction with segmentation segment type to same road difference is belonged to according to the lane quantity detected by segmentation section
The lane quantity of section compares, and judges whether the road track quantity changes.
For example, in same road, successive segmentation section x2-x15Classification results analyzed, in fact it could happen that intersection
The inconsistent situation of the lane detection result of each segmentation section, compares each segmentation section and corresponds to FCD distribution characteristics, it is found that the road
The data of mouthful Floating Car dispersion of distribution are more discrete and the variation presentation of track cover width value gradually decreases trend, in conjunction with segmentation
Section number of track-lines amount calculates as a result, obtain at crossing that there are road widenings, that is, increases a lane.
When it is implemented, computer software technology, which can be used, in above technical scheme realizes automatic running process.
Technical solution for a better understanding of the present invention is really floated with the method for the embodiment of the present invention below
Car data cleaning experiment.
1, true floating car data introduction
Track data is Wuhan City's in August, 2013 totally 15 days taxi GPS tracks, and track data includes vehicle ID, GPS
The information such as time, GPS longitude and latitude, track sample frequency are 40s.
2, the analysis of intersection lane change type and its corresponding Floating Car track characteristic mutation analysis
According to city intersection planning and design specification, in conjunction with current physical presence grade crossing image, often at present
The intersection road track seen is (with import lane[58]For) quantity situation of change includes: that intersection lane quantity is unchanged
Change, broaden 1 lane or a plurality of lane, crossroad on the left of 1 lane of broadening or a plurality of lane, intersection on the right side of intersection
Mouth bilateral broadens 2 lanes or a plurality of lane.
Since road junction adds broadening lane, change compared to the road middle section road track quantity for belonging to road,
The track characteristic of Floating Car also changes correspondingly in corresponding road surface, including the variation of track geometrical characteristic and track motion feature become
Change.In terms of the geometrical characteristic of track, Floating Car is generally referred in the process of moving, vehicle is passed through roadway widening by road middle section
During transition reaches intersection position, since road junction driveway travel directions limit, the floating on road
Vehicle is according to guide is turned to, and in broadening transition changing Lane, trajectory line also changes therewith, so floating wheel paths are in road surface
Cover width it is also broadening therewith, while to the track in lane each on road surface be evenly distributed density reduction.It is moved in track
Characteristic aspect, due to floating vehicle travelling lane-change, the instantaneous driving direction of Floating Car also changes, and is mainly manifested in Floating Car
Instantaneous driving direction and road surface track traveling principal direction angle increase；Simultaneously as intersection traffic signal is converted, so that
Near the stop line of intersection, Floating Car is due to parking behavior, so that Floating Car is in halted state, floating vehicle speed and side
It is 0 to angle, this kind of situation is mainly manifested in, the floating car data covered in road junction lane, compared in road
The floating car data in road middle section, floating vehicle speed and Floating Car angular separation change.
3, the crossing broadening lane based on the method for the present invention calculates
It plans that specification [18] divide region to same road according to urban road junction, summarizes and summarize road difference area
The FCD spatial distribution characteristic covered on domain is carried out division region to road using city road network polar plot, is averagely filtered using clipping
Then lane quantity known to a part and FCD are covered shape by the method excluding gross error of wave and the characteristic parameter for obtaining preference data
The section of state constructs classifier as training sample, lane quantity type is confirmed using GBDT, finally to road different zones
Lane is subject to comparative analysis, and then efficiently detects road junction lane information, and especially broadening lane quantity in crossing is visited
It surveys.
Lane quantity based on FCD, which obtains, uses GBDT model, promotes decision tree according to the gradient that document [18,19] propose
Classification method, essentially by the distribution of floating car data on analysis road surface, into the classification number to speculate GBDT model,
With the GBDT model superiority and inferiority after the combination of this comparison different classifications, when Selection Model minimum, corresponding classification number was as this section of road
Corresponding number of lanes.
4, true crossing broadening lane experiment with computing result and evaluation
Track data is Wuhan City's in August, 2013 totally 15 days taxi GPS tracks, and track data includes vehicle ID, GPS
The information such as time, GPS longitude and latitude, track sample frequency are 40s.Road data is Wuhan City's standard road vector number in 2012
According to.Since the historical data that FCD track data is 2013 is studied herein for the accuracy for verifying this experiment with Wuhan City's force
More typical four roads in prosperous area amount to 21 grade crossings, and do not construct during 2013 to 2014
Experimental study is unfolded as training sample in the road of construction.Road data is stored using ArcGIS document data bank, with .xls text
Part is as intermediate conversion file, and using the filtering experiments of Matlab software FCD, GBDT sorting algorithm is using scikit-
Learn is tested in library.
Three kinds of methods are respectively adopted and extract intersection and non-crossing road junction roadway number, compare real image, calculate each side
The accuracy of method, as shown in table 1.
1 number of track-lines amount of table calculates quantitative assessment
It uses the overall accuracy of context of methods detection road junction lane quantity for 81.86%, detects non-crossing road
The overall accuracy of mouth lane quantity is 83.89%, is better than other two methods.It can be summarized as at following 2 points by analyzing reason:
The space characteristics of different zones FCD are targetedly studied in the road ①Jiang k-path partition intersection region and non-crossing crossing region, point
Not Gou Jian GBDT disaggregated model, improve to section FCD to be measured simulate effect, improve road junction roadway number judgement precision；②
The attribute data for implying GPS track information in FCD is made full use of, v is added, ɑ broadens gradual change as characteristic of division, to road track
In the process, when judging lane quantity by unitary variant d, the biggish defect of deviation is made up, to improve road track quantity
The overall precision of judgement.
Specific embodiment described herein is only an example for the spirit of the invention.The neck of technology belonging to the present invention
The technical staff in domain can make various modifications or additions to the described embodiments or replace by a similar method
In generation, however, it does not deviate from the spirit of the invention or beyond the scope of the appended claims.
Claims (4)
1. a kind of crossing for promoting decision tree based on gradient broadens lane detection method, which comprises the following steps:
Step 1, data input, including input floating car data and urban road polar plot carry out data prediction, reject and float
Then floating car data is divided into the Floating Car at the floating car data and non-crossing mouth of intersection by the shift point in wheel paths
Data, then the road covered respectively according to track are equidistantly divided along road direction, obtain several compartmenteds as basic
Study unit；
Step 2, signature analysis, including intersection lane broaden Variations, and corresponding with the variation of intersection lane
The analysis of FCD track characteristic；
Step 3, feature selecting, including selecting 4 characteristic parameters for constructing basic classification device and the calculating of number of track-lines amount, respectively
The dispersion of distribution, distribution density, the angular separation of FCD and the speed that are floating car data point on road cross section；
Step 4, basic classification device is constructed, including the Run-time scenario according to floating wheel paths, constructs the vehicle based on FCD probability density
The basic classification device of road quantity；
Step 5, the lane for promoting decision tree based on gradient calculates, including the use of floating car data, referring to basic classification device, according to
Gradient promotes Decision-Tree Method, and when FCD is divided different lane quantity by selection, comentropy is minimum and the driveway partition is in cross
The corresponding classification number of the state that is optimally distributed on section is as the corresponding number of lanes of this section of road；
Step 6, broadening lane determines, including on the same road of comparison, number of track-lines on the corresponding road section of difference research unit
Situation of change is measured, judges crossing with the presence or absence of broadening lane.
2. the crossing for promoting decision tree based on gradient according to claim 1 broadens lane detection method, it is characterised in that: step
In rapid 4, using the floating car data covered on the section of known lane quantity as training sample, according to descriptive statistic side
Method counts the average value of 4 characteristic parameters of floating car data covered on the section of different lane quantity respectively, realizes building base
This classifier.
3. the crossing according to claim 1 or claim 2 for promoting decision tree based on gradient broadens lane detection method, feature exists
In: the dispersion of distribution of the floating car data point on road cross section is arrived by calculating each floating car data point in segmentation section
The Euclidean distance value of track centers obtains.
4. the crossing according to claim 1 or claim 2 for promoting decision tree based on gradient broadens lane detection method, feature exists
In: distribution density extracting mode of the floating car data point on road cross section is as follows,
The trace centerline horizontal direction along segmentation section is counted by floating car data coverage area by several sections are equidistantly divided into
It calculates points amount in track in each section and divides the ratio between total quantity in section, obtain the track dot density in each section；By any floating
The track dot density in section where the corresponding tracing point of car data point, as the distribution of the floating car data point on road cross section
The value of density.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810877633.9A CN108961758B (en) | 2018-08-03 | 2018-08-03 | Road junction widening lane detection method based on gradient lifting decision tree |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810877633.9A CN108961758B (en) | 2018-08-03 | 2018-08-03 | Road junction widening lane detection method based on gradient lifting decision tree |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108961758A true CN108961758A (en) | 2018-12-07 |
| CN108961758B CN108961758B (en) | 2021-02-02 |
Family
ID=64467325
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810877633.9A Active CN108961758B (en) | 2018-08-03 | 2018-08-03 | Road junction widening lane detection method based on gradient lifting decision tree |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108961758B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110414067A (en) * | 2019-07-01 | 2019-11-05 | 东南大学 | A method of considering that the cell inside and outside road of traffic safety is connected design |
| CN110598747A (en) * | 2019-08-13 | 2019-12-20 | 广东工业大学 | Road classification method based on self-adaptive K-means clustering algorithm |
| CN112489419A (en) * | 2020-10-28 | 2021-03-12 | 华为技术有限公司 | Method and device for determining road capacity and storage medium |
| CN112699190A (en) * | 2019-10-23 | 2021-04-23 | 合肥四维图新科技有限公司 | Map data detection method, device, system and storage medium |
| CN113052265A (en) * | 2021-04-26 | 2021-06-29 | 上海海事大学 | Moving object track simplification algorithm based on feature selection |
| CN113052118A (en) * | 2021-04-07 | 2021-06-29 | 上海浩方信息技术有限公司 | Method, system, device, processor and storage medium for realizing scene change video analysis and detection based on high-speed dome camera |
| US11580850B2 (en) | 2020-12-09 | 2023-02-14 | Here Global B.V. | Method, apparatus and computer program product for determining lane status confidence indicators using probe data |
| CN117077042A (en) * | 2023-10-17 | 2023-11-17 | 北京鑫贝诚科技有限公司 | Rural level crossing safety early warning method and system |
| US11892317B2 (en) | 2021-12-20 | 2024-02-06 | Here Global B.V. | Automatic detection of segment width narrowing using probe data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1498694A2 (en) * | 2003-07-16 | 2005-01-19 | Navigation Technologies Corporation | Method of representing road lanes |
| US20100312527A1 (en) * | 2003-06-19 | 2010-12-09 | Michael Weiland | Method of representing road lanes |
| CN103031790A (en) * | 2011-10-07 | 2013-04-10 | 徐晓广 | Road facility capable of enabling vehicles on crossing road of city not to intersecting |
| CN104700617A (en) * | 2015-04-02 | 2015-06-10 | 武汉大学 | High-precision lane information extracting method based on low-precision GPS track data |
-
2018
- 2018-08-03 CN CN201810877633.9A patent/CN108961758B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100312527A1 (en) * | 2003-06-19 | 2010-12-09 | Michael Weiland | Method of representing road lanes |
| US20170045370A1 (en) * | 2003-06-19 | 2017-02-16 | Here Global B.V. | Method of representing road lanes |
| EP1498694A2 (en) * | 2003-07-16 | 2005-01-19 | Navigation Technologies Corporation | Method of representing road lanes |
| CN103031790A (en) * | 2011-10-07 | 2013-04-10 | 徐晓广 | Road facility capable of enabling vehicles on crossing road of city not to intersecting |
| CN104700617A (en) * | 2015-04-02 | 2015-06-10 | 武汉大学 | High-precision lane information extracting method based on low-precision GPS track data |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110414067A (en) * | 2019-07-01 | 2019-11-05 | 东南大学 | A method of considering that the cell inside and outside road of traffic safety is connected design |
| CN110414067B (en) * | 2019-07-01 | 2023-07-07 | 东南大学 | Method for designing intra-district and external road connection considering traffic safety |
| CN110598747A (en) * | 2019-08-13 | 2019-12-20 | 广东工业大学 | Road classification method based on self-adaptive K-means clustering algorithm |
| CN112699190A (en) * | 2019-10-23 | 2021-04-23 | 合肥四维图新科技有限公司 | Map data detection method, device, system and storage medium |
| CN112489419B (en) * | 2020-10-28 | 2022-04-26 | 华为技术有限公司 | Method and device for determining road capacity and storage medium |
| CN112489419A (en) * | 2020-10-28 | 2021-03-12 | 华为技术有限公司 | Method and device for determining road capacity and storage medium |
| US11580850B2 (en) | 2020-12-09 | 2023-02-14 | Here Global B.V. | Method, apparatus and computer program product for determining lane status confidence indicators using probe data |
| CN113052118A (en) * | 2021-04-07 | 2021-06-29 | 上海浩方信息技术有限公司 | Method, system, device, processor and storage medium for realizing scene change video analysis and detection based on high-speed dome camera |
| CN113052265A (en) * | 2021-04-26 | 2021-06-29 | 上海海事大学 | Moving object track simplification algorithm based on feature selection |
| CN113052265B (en) * | 2021-04-26 | 2024-04-05 | 上海海事大学 | Moving object track simplifying method based on feature selection |
| US11892317B2 (en) | 2021-12-20 | 2024-02-06 | Here Global B.V. | Automatic detection of segment width narrowing using probe data |
| CN117077042A (en) * | 2023-10-17 | 2023-11-17 | 北京鑫贝诚科技有限公司 | Rural level crossing safety early warning method and system |
| CN117077042B (en) * | 2023-10-17 | 2024-01-09 | 北京鑫贝诚科技有限公司 | Rural level crossing safety early warning method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108961758B (en) | 2021-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Automatic intersection and traffic rule detection by mining motor-vehicle GPS trajectories | |
| CN108961758A (en) | A kind of crossing broadening lane detection method promoting decision tree based on gradient | |
| CN107766808B (en) | Method and system for clustering moving tracks of vehicle objects in road network space | |
| CN109670277B (en) | Travel time prediction method based on multi-mode data fusion and multi-model integration | |
| CN104700617B (en) | Based on the high precision track information extracting method of low precision GPS track data | |
| CN110298500B (en) | Urban traffic track data set generation method based on taxi data and urban road network | |
| CN108920481B (en) | Road network reconstruction method and system based on mobile phone positioning data | |
| CN110909788B (en) | Statistical clustering-based road intersection position identification method in track data | |
| CN102102992B (en) | Multistage network division-based preliminary screening method for matched roads and map matching system | |
| CN109410586A (en) | A kind of Traffic State Detection Method based on multivariate data fusion | |
| Patnaik et al. | Divisive Analysis (DIANA) of hierarchical clustering and GPS data for level of service criteria of urban streets | |
| CN104331422A (en) | Road section type presumption method | |
| CN103839409A (en) | Traffic flow state judgment method based on multiple-cross-section vision sensing clustering analysis | |
| CN112991743B (en) | Real-time traffic risk AI prediction method based on driving path and system thereof | |
| CN110851741A (en) | Taxi passenger carrying hot spot identification recommendation algorithm | |
| CN116542709A (en) | Electric vehicle charging station planning analysis method based on traffic situation awareness | |
| CN109489679A (en) | A kind of arrival time calculation method in guidance path | |
| Yu et al. | Identifying traffic clusters in urban networks based on graph theory using license plate recognition data | |
| Zhang et al. | Recent developments in traffic flow modeling using macroscopic fundamental diagram | |
| Bhuyan | Defining Level of Service Criteria for Urban Streets in Indian Context | |
| CN106949897B (en) | Method and device for generating roads in map | |
| CN113888867B (en) | Parking space recommendation method and system based on LSTM (least squares) position prediction | |
| Mohapatra et al. | Genetic algorithm fuzzy clustering using GPS data for defining level of service criteria of urban streets | |
| CN109147322B (en) | Multi-source data self-adaptive fusion method in urban traffic big data processing | |
| Raj et al. | Evaluation of perception and nonperception based approaches for modeling urban road level of service |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20231225 Address after: 230000 floor 1, building 2, phase I, e-commerce Park, Jinggang Road, Shushan Economic Development Zone, Hefei City, Anhui Province Patentee after: Dragon totem Technology (Hefei) Co.,Ltd. Address before: 330013 No.418, Guanglan Avenue, Changbei Economic and Technological Development Zone, Nanchang City, Jiangxi Province Patentee before: EAST CHINA INSTITUTE OF TECHNOLOGY |
|
| TR01 | Transfer of patent right |