CN107533630A - For the real time machine vision of remote sense and wagon control and put cloud analysis - Google Patents
For the real time machine vision of remote sense and wagon control and put cloud analysis Download PDFInfo
- Publication number
- CN107533630A CN107533630A CN201680006431.6A CN201680006431A CN107533630A CN 107533630 A CN107533630 A CN 107533630A CN 201680006431 A CN201680006431 A CN 201680006431A CN 107533630 A CN107533630 A CN 107533630A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- data
- assets
- map
- information
- 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
- 238000004458 analytical method Methods 0.000 title claims description 28
- 238000000034 method Methods 0.000 claims abstract description 72
- 238000012545 processing Methods 0.000 claims abstract description 51
- 230000007246 mechanism Effects 0.000 claims abstract description 47
- 238000013500 data storage Methods 0.000 claims abstract description 17
- 238000007405 data analysis Methods 0.000 claims abstract description 14
- 238000004891 communication Methods 0.000 claims description 36
- 238000001514 detection method Methods 0.000 claims description 24
- 238000007906 compression Methods 0.000 claims description 18
- 230000006835 compression Effects 0.000 claims description 18
- 238000012795 verification Methods 0.000 claims description 18
- 239000013598 vector Substances 0.000 claims description 17
- 238000003860 storage Methods 0.000 claims description 9
- 238000013480 data collection Methods 0.000 claims description 8
- 230000004888 barrier function Effects 0.000 claims description 6
- 238000004519 manufacturing process Methods 0.000 claims description 6
- 238000012550 audit Methods 0.000 claims description 5
- 230000006837 decompression Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims 3
- 230000004075 alteration Effects 0.000 claims 2
- 238000010801 machine learning Methods 0.000 abstract description 11
- 238000000605 extraction Methods 0.000 abstract description 9
- 238000005457 optimization Methods 0.000 abstract description 4
- 230000004438 eyesight Effects 0.000 description 41
- 230000008569 process Effects 0.000 description 28
- 238000004422 calculation algorithm Methods 0.000 description 24
- 239000000306 component Substances 0.000 description 17
- 238000010586 diagram Methods 0.000 description 14
- 238000012549 training Methods 0.000 description 11
- 238000005259 measurement Methods 0.000 description 10
- 230000008859 change Effects 0.000 description 9
- 238000003908 quality control method Methods 0.000 description 9
- 230000009471 action Effects 0.000 description 8
- 230000007613 environmental effect Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 238000000060 site-specific infrared dichroism spectroscopy Methods 0.000 description 8
- 230000036541 health Effects 0.000 description 7
- 238000006243 chemical reaction Methods 0.000 description 6
- 238000012790 confirmation Methods 0.000 description 6
- 230000000875 corresponding effect Effects 0.000 description 6
- 238000007689 inspection Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 230000011664 signaling Effects 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 4
- 230000001276 controlling effect Effects 0.000 description 4
- 230000001419 dependent effect Effects 0.000 description 4
- 238000009826 distribution Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000033001 locomotion Effects 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000009467 reduction Effects 0.000 description 4
- 238000005096 rolling process Methods 0.000 description 4
- 238000012800 visualization Methods 0.000 description 4
- 230000006399 behavior Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 230000002708 enhancing effect Effects 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 238000006116 polymerization reaction Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 241001669679 Eleotris Species 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000013144 data compression Methods 0.000 description 2
- 238000013075 data extraction Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 239000000446 fuel Substances 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 229910052742 iron Inorganic materials 0.000 description 2
- 230000003137 locomotive effect Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000007935 neutral effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000000704 physical effect Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000001429 visible spectrum Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- YPFNIPKMNMDDDB-UHFFFAOYSA-K 2-[2-[bis(carboxylatomethyl)amino]ethyl-(2-hydroxyethyl)amino]acetate;iron(3+) Chemical compound [Fe+3].OCCN(CC([O-])=O)CCN(CC([O-])=O)CC([O-])=O YPFNIPKMNMDDDB-UHFFFAOYSA-K 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 240000002853 Nelumbo nucifera Species 0.000 description 1
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 1
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 1
- 244000131316 Panax pseudoginseng Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 125000002015 acyclic group Chemical group 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000011049 filling Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 230000007257 malfunction Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000004297 night vision Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 235000019353 potassium silicate Nutrition 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 230000003449 preventive effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000001681 protective effect Effects 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000029058 respiratory gaseous exchange Effects 0.000 description 1
- 238000007665 sagging Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- NTHWMYGWWRZVTN-UHFFFAOYSA-N sodium silicate Chemical compound [Na+].[Na+].[O-][Si]([O-])=O NTHWMYGWWRZVTN-UHFFFAOYSA-N 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000013517 stratification Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 238000004804 winding Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L23/00—Control, warning or like safety means along the route or between vehicles or trains
- B61L23/34—Control, warning or like safety means along the route or between vehicles or trains for indicating the distance between vehicles or trains by the transmission of signals therebetween
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L23/00—Control, warning or like safety means along the route or between vehicles or trains
- B61L23/04—Control, warning or like safety means along the route or between vehicles or trains for monitoring the mechanical state of the route
- B61L23/041—Obstacle detection
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L25/00—Recording or indicating positions or identities of vehicles or trains or setting of track apparatus
- B61L25/02—Indicating or recording positions or identities of vehicles or trains
- B61L25/025—Absolute localisation, e.g. providing geodetic coordinates
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L25/00—Recording or indicating positions or identities of vehicles or trains or setting of track apparatus
- B61L25/02—Indicating or recording positions or identities of vehicles or trains
- B61L25/026—Relative localisation, e.g. using odometer
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L27/00—Central railway traffic control systems; Trackside control; Communication systems specially adapted therefor
- B61L27/04—Automatic systems, e.g. controlled by train; Change-over to manual control
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L3/00—Devices along the route for controlling devices on the vehicle or train, e.g. to release brake or to operate a warning signal
- B61L3/02—Devices along the route for controlling devices on the vehicle or train, e.g. to release brake or to operate a warning signal at selected places along the route, e.g. intermittent control simultaneous mechanical and electrical control
- B61L3/08—Devices along the route for controlling devices on the vehicle or train, e.g. to release brake or to operate a warning signal at selected places along the route, e.g. intermittent control simultaneous mechanical and electrical control controlling electrically
- B61L3/12—Devices along the route for controlling devices on the vehicle or train, e.g. to release brake or to operate a warning signal at selected places along the route, e.g. intermittent control simultaneous mechanical and electrical control controlling electrically using magnetic or electrostatic induction; using radio waves
- B61L3/127—Devices along the route for controlling devices on the vehicle or train, e.g. to release brake or to operate a warning signal at selected places along the route, e.g. intermittent control simultaneous mechanical and electrical control controlling electrically using magnetic or electrostatic induction; using radio waves for remote control of locomotives
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
- G01S17/931—Lidar systems specially adapted for specific applications for anti-collision purposes of land vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/59—Context or environment of the image inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L2205/00—Communication or navigation systems for railway traffic
- B61L2205/04—Satellite based navigation systems, e.g. global positioning system [GPS]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/02—Recognising information on displays, dials, clocks
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Electromagnetism (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Train Traffic Observation, Control, And Security (AREA)
- Electric Propulsion And Braking For Vehicles (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The method and apparatus for real time machine vision and cloud data analysis are provided, for remote sense and wagon control.Cloud data can be analyzed via expansible centralized cloud computing system, for withdrawal of assets information and generative semantics map.Data set is finely divided for streaming to distributed processing unit (1240) by data storage/preprocessor (1220), and is operated via data analysis mechanism (1250).The output of processing unit (1240) is polymerize by map generator (1230).Machine learning component can optimize data analysis mechanism, to improve assets and the feature extraction from sensing data.The data analysis mechanism of optimization can be downloaded to vehicle to be used to analyze vehicle sensor data in onboard system.Semantic map datum can locally use together with onboard sensor in vehicle, to obtain accurate vehicle location and input to vehicle offer to control system.
Description
The cross reference of related application
This application claims entitled " the A Scalable Approach To Point- submitted on January 20th, 2015
Cloud Data Processing for Railroad Asset Location and Health Monitoring (are used for
Railroad assets position and health monitoring Point Cloud Processing expandable method) " U.S. Provisional Patent Application No.62/
105,696 senior interest, the entire disclosure of which are incorporated by reference into herein.
Background technology
In some different fields, the long-range sense based on machine being automatically positioned with vehicle home environment of mobile vehicle
Survey becomes more and more important.One such field is exactly motor vehicle transport.In recent years, many automobiles and truck are implemented vehicle-mounted complete
Ball alignment system (GPS) receiver and navigation system, so as to guide driver using gps data.However, due to motor vehicle system
Make business and pursue and implement more advanced navigation automation, such as autonomous driving feature, so the alignment system based on GPS may be both
Sufficiently exact vehicle location can not be provided, they do not allow to sense the home environment of vehicle in real time yet.Accordingly, it is possible to
It is expected complementarity sensing system, and very detailed infrastructure and terrestrial reference map, three-dimensional semantic map may be included.
It may be desirable to vehicle location, being sensed to home environment and another application of three-dimensional semantic map is to arrange
In the operation of car.US Congress has passed through american railway safety improvement bill (Rail Safety Improvement in 2008
Act) to ensure to monitor all trains in real time, so as to enable " active Train Control " (PTC, Positive Train
Control, train active control).All trains of the laws and regulations requirement report their positional information so that real-time tracing owns
Train movement.It is required that PTC can be operated in having both signal region and no signal (dark, no to broadcast, be dim) region.
In order to realize this milestone, many companies have attempted to implement various PTC systems.The problem of reappearing is current
PTC system only can train pass through along railway line set roadside ask answer machine or signal station when following train, cause to grasp
Work person does not know state of the train by the road between signal.Therefore, the distance between continuous physics wayside signal infrastructure
Minimum safe distance (spacing distance (headway, interval time)) needed for determining between train.Due to along railway network
Length structure PTC infrastructure and the cost and complexity that are maintained, current signaling infrastructure also limit
Dispose the scope of wayside signal facility.The current train the last time that is used to detect, the method for process passed through near detector by the road
The positional information answered by inquiry between machine lacks.
Some companies are further utilized along the radio tower of the rail network of operator to create between train
Virtual signal, avoid the demand to wayside signal facility.In order to carry out radio communication, radio tower still needs deployment letter
Make facility.However, for reliable positional information, it is necessary to dispose extra inquiry along the track of train and answer machine, with reliably true
Determine the position of train and track that train is currently just taking.
One example of PTC system in use is European Train Control System (ETCS), and it depends on by track and set
Apply and pair information relevant with signaling is made a response the control device of installation ON TRAINS.The system depends critically upon not
In the infrastructure that the U.S. or developing country dispose.
It is required that the scheme for minimal disposing wayside signal facility is active for being established in the whole U.S. and developing country
Train Control is beneficial.Millions of transponders is disposed per 1-15km along track --- for detecting and transmitting depositing for train
Machine is answered in inquiry in the position with train --- less effectively, because transponder can be by environmental aspect, the negative shadow of stealing
Ring, and transponder needs General Maintenance, and the data gathered may not used in real time.In view of throughout the whole railway network
Network PTC utilizes the cost of transponder, and it is not expansible scheme only to obtain position data by facility by track.In addition, train
Control and security system can not depend only on global positioning system (GPS), because it can not be sufficiently accurately in track
Between make a distinction, thus need wayside signal to carry out position correction.
With the development of autonomous driving, Train Control and other vehicle operating systems, pass through systems described below and side
Method can solve the challenge of these and other.
The content of the invention
According to one side disclosed herein, describe for positioning and/or controlling vehicle such as train or motor vehicle
System and method.Can be installed on vehicle can include NI Vision Builder for Automated Inspection such as LiDAR home environment sensor.May be used also
To provide the first geographical position of vehicle including gps receiver.Remote data base and processor are stored and handled from multiple sources
The data of collection, and on-vehicle vehicle processor is downloaded and the running of mobile vehicle, the data that safety and/or control are related.This
The data of ground environmental sensor generation description surrounding environment, the cloud data such as generated by LiDAR sensors.The number gathered
According to can be handled in processing locality, on vehicle, or remote data system is uploaded to be stored, be handled and be analyzed.
Analysis institution (vehicle-mounted and/or implement in remote data system) can operate to the data gathered, with from biography
Sensor extracting data information, the mark of the object such as in home environment and position.
One exemplary of system described herein includes being arranged on the Hardware Subdivision on railway or on other vehicles
Part, remote data base and analysis component, the analysis component to handle gathered with the information-related of transportation system
Data, including mobile and stationary vehicle, infrastructure and transportation route (such as rail or road) situation.System can such as lead to
Cross the position for the object that will be detected in the onboard sensor of vehicle and the known location of object is compared to estimate exactly
The exact position that meter vehicle is advanced along transportation route.Other attribute on example components can be described in detail herein, and
And including herein below:
Hardware:In order to inform the movement of vehicle safely, including:In railway applications, in addition to other features, car is identified
Track, barrier where advancing, the health of track and rail system;And in road vehicle application, except other features
In addition, track, the texture and health, the mark of neighbouring assets of road where vehicle is advanced.
Remote data base:Comprising the information on assets, and it can be remotely inquired about to obtain extra assets
Information.
Database with assets information is overall (population):Method is included by traveling vehicle gathered data itself,
Or by other vehicle (dual-purpose vehicle for highway and railway, track inspection car, aircraft, mobile Mapping Platform etc.) collection numbers
According to.Then the data are handled to generate assets information (among other information, position, feature, road/track health).
Data analysis mechanism:By some data and information flow (Tathagata from sensor, database, roadside unit, vehicle letter
Breath bus etc.) it is fused together to produce the accurate estimation to track, track ID or other telltale marks.
According to provided herein is text and accompanying drawing, the these and other aspects of disclosure will be apparent.
Brief description of the drawings
Exemplary is will now further be described with reference to the drawings, identical reference represents identical in accompanying drawing
Element, and:
Fig. 1 is the representative flow diagram of train control system;
Fig. 2 is the representative flow diagram of the vehicle-mounted ecosystem;
Fig. 3 is the representative flow diagram for obtaining positional information;
Fig. 4 is the example plot that train infers signal lamp state;
Fig. 5 is can be as example plot of the feedback for the various interfaces of attendant;
Fig. 6 is for obtaining by the track ID of train occupation representative flow diagram;
Fig. 7 is the representative flow diagram for describing track ID algorithms;
Fig. 8 is the representative flow diagram for describing signal lamp (signal, signalling means) state algorithm;
Fig. 9 is the representative flow diagram for describing sensing and feedback;And
Figure 10 is the representative flow diagram for the image mosaic technology of related track positioning.
Figure 11 A and Figure 11 B are a flow charts for cloud analysis process.
Figure 12 is the schematic block diagram for the device of a cloud analysis.
Figure 13 is the flow chart for analyzing the process of cloud data.
Figure 14 is the another flow chart for analyzing the process of cloud data.
Figure 15 is to show point cloud chart block (tile) size and Density Distribution in exemplary dots cloud surveys (survey)
Chart.
Figure 16 is points cloud processing cluster (cluster) schematic block diagram.
Figure 17 is the plot of the characteristic for the compression mechanism that cloud data can be used.
Figure 18 is the plot of the characteristic for the compression mechanism that cloud data can be used.
Figure 19 is the plot of the characteristic for the compression mechanism that cloud data can be used.
Figure 20 is the flow chart of the process for track detecting.
Figure 21 is the visualization of the point cloud section of the rail information with extraction.
Figure 22 is the histogram of exemplary dots cloud section point cloud intensity rank.
Figure 23 is the visualization of track detecting mechanism output.
Figure 24 is the schematic block diagram using the map generation system of supervision machine study.
Figure 25 is the schematic block diagram for the run time system of motor vehicle positioning, motor vehicle control and map examination & verification.
Embodiment
According to an embodiment, there is provided for determining one or more mobile vehicles such as train or autonomous land vehicle
Position and independent of the transponder/inquiry method and apparatus of answering machine for spreading all over running environment distribution for accurate position data.
Implementation of some of this embodiment based on train be herein referred to as BVRVB-PTC, PTC vision system or
NI Vision Builder for Automated Inspection.
Also disclose using the position data to optimize wagon control and operation such as operation of the train in rail system
Scheme.Railway embodiment can be merged using a series of sensors and Data fusion technique is come with improved accuracy and can
Orbital position is obtained by property.This embodiment can be with:For Braking mode of the train due to violating red light in orbit;With
In optimizing fuel based on region, make train speed synchronous to avoid red light, collision avoidance system;And for being not only train
But also the protective for having the gravel bottom (substrate) below track, rail and track is safeguarded.Some embodiments can
To use back-end processing and memory unit, for keeping tracking assets position and health and fitness information (can be accessed by mobile vehicle
Or can be accessed by railroad operator by reporting).
Except positioning, for the embodiment of autonomous driving, it may be desirable to utilize very detailed infrastructure and terrestrial reference
Map.These maps can be used for the magnitude of traffic flow guided in real world, and is planned for vehicle from source point and advance to purpose
The route on ground.The three-dimensional nature --- except they are in addition to representing in terms of the accuracy of physical world --- of map helps vehicle
Expect the event beyond their sensing range, by their sensor recessed (foveate) to assets interested, and on
Terrestrial reference positioning vehicle.By the way that very detailed three-dimensional (semanteme) map is used for into pseudo- static assets, the resource of vehicle be released with
Observe the dynamic object of vehicle periphery.
PTC vision systems can include processing communication module, image-capturing apparatus, image processing equipment, computing device,
The data aggregate platform and inertial sensor (including onboard sensor and position sensor) docked with train signal bus.
Fig. 1 shows the exemplary flow operation of train control system.In the step s 100, train experience normal operation.
In step S105, train status is obtained from data aggregate platform is (described below).In step s 110, (refine) is refined
Train position.In step sl 15, semaphore lamp state is identified from home environment sensor information.In the step s 120,
Using feedback.Train speed (step S125) can be adjusted, alarm and/or notice (step S130) can be sent.It can hereafter retouch
State on the further detail below of each in these steps.
Reference picture 2, PTC vision systems can include one or more of following:Data aggregate platform (DAP) 215, regard
Feel device (VA) 230, active Train Control computer (PTCC) 210, man-machine interface (HMI) 205, gps receiver 225 and
The vehicle communication device (VCD) 220 generally to be communicated via LAN or WAN communication network 240.
Part (such as VCD, HMI, PTCC, VA, DAP, GPS) is desirably integrated into single part or can be inherently
It is modular, and can be virtual software or physical hardware devices.Each part in PTC vision systems can have it
The power supply of itself shares power supply with PTCC.Power supply for the part in PTC vision systems can be included for power-off
The not interruption member of (power outage).
PTCC modules are maintained in the state for the information passed through between the module of PTC vision systems.PTCC and HMI, VA, VCD,
GPS and DAP communications.Communication can include providing information (such as data) and/or receive information.The ecosystem any module it
Between interface (e.g., bus, connector) any conventional interface can be included.The module of the ecosystem can use any routine
Communication protocol communicates with one another, communicated with human operator who and/or with third party (such as another train, another attendant, another train
Operator) communication.Communication can be realized via wiredly and/or wirelessly communication link (such as channel).
Any conventional treatment circuit can be used --- including microprocessor, computer, signal processor, memory and/
Or bus --- to implement PTCC.PTCC can perform any calculating for the function of being adapted for carrying out PTC vision systems.
HMI module can be from PTCC module receive informations.The information received by HMI module can include:Geographical position is (such as
GPS dimension & longitude coordinates);Time;The speed of recommendation;Directionality course (such as orientation);Track ID;Neighbour on the same track
The distance between nearly train/spacing distance;The distance between neighbouring train on adjacent orbit/spacing distance;Website interested,
Including next website, website before or the website between starting point and destination;The orbital segment currently used for train
Virtually or physically arm plate state;Orbital segment for the orbital segment that will be reached in the route of train and before virtual or
The state of physics arm plate;And the state of the virtually or physically arm plate of the orbital segment for sharing track interlocking with current orbit.
HMI module can provide information to PTCC modules.The letter from operator can be included by being supplied to PTCC information
Breath and/or request.HMI can provide information to preceding processing (e.g., formatting, reduction, the tune of operator or PTCC modules
It is whole, related) information.The information for being supplied to PTCC modules by HMI can include:Attendant is ordered to train deceleration;Attendant please
Ask and ignore some parameters (such as constraint of velocity);Confirmation of the attendant to message (e.g., failure, status information);Attendant is to extra
The request of information (e.g., diagnostic program, the accident along railroad track or other point-of-interests along railroad track);And
Any other information of interest relevant with the train operation of attendant.
HMI provides user interface (such as GUI) to human user (e.g., attendant, operator).Human user can operate
The control device (e.g., button, control-rod, knob, touch-screen, keyboard) of HMI module, to provide information or request to HMI module
Information from vision system.Operator can wear the user interface of HMI module.User interface can via tactile manipulation,
Wire communication and/or radio communication communicate with HMI module.The information provided a user by HMI module can include:The speed of recommendation
Degree, current speed, efficiency value or index, driver profile, wayside signal lamp state, website interested, the map of inertia measurement
View, failure message, alarm, train operator interfaces for activating train head controlling device and for confirmation message or notice
Train operator interfaces.
VCD modules perform communication (as wired, wireless).VCD modules enable PTC vision systems with ON TRAINS
Not other equipment communication ON TRAINS.VCD modules can provide wide area network (" WAN ") communication and/or LAN (" LAN ")
Communication.Any Conventional communication techniques and/or agreement (e.g., honeycomb, satellite, dedicated channel) can be used to perform WAN communication.Can be with
Use any Conventional communication techniques and/or agreement (e.g., Ethernet, WiFi, bluetooth, wireless HART, low-power consumption WiFi, low energy consumption
Bluetooth, optical fiber, IEEE 802.15.4e) perform LAN communication.The one of frequency used in being suitable for and/or agreement can be used
Kind or a variety of antennas perform radio communication.
VCD modules can be from PTCC module receive informations.VCD can transmit the information received from PTCC modules.Information
General headquarters' (such as central station), roadside facility, individual and/or other trains can be transferred to.Information from PTCC modules can be with
Including:It is addressed to the packet of other trains;Shared back-end server is addressed to inform the data of train position to operator
Bag;It is addressed to the packet of roadside facility;Roadside personnel are addressed to transmit the packet of train position;Any node-to-node
Arbitrary Loads;And it is addressed to the packet of the third party audience of PTC vision systems.
VCD modules can also provide information to PTCC modules.VCD can transmit any source of information from VCD to it
Receive information.The information provided from VCD to PTCC can include:The packet being addressed from other trains;After sharing
End server rise be addressed to give the packet of feedback to attendant or train;The number being addressed from the facility of roadside
According to bag;Be addressed from the personnel of roadside to transmit the packet of personnel positions;The Arbitrary Loads of any node-to-node;With
And the packet being addressed from the third party audience of PTC vision systems.
GPS module can include conventional Global alignment system (" GPS ") receiver.GPS module receives from gps satellite to be believed
Number, and determine the geographical position of receiver and time (such as UTC time) using the information provided by these signals.GPS module
It can include being used for one or more antennas from satellite received signal.Antenna can be arranged to reduction and/or detection multichannel
Footpath signal and/or mistake.GPS module can safeguard the historical record of geographical position and/or time.GPS module can determine to arrange
The speed and direction that garage enters.GPS module can receive control information (e.g., WAAS, difference) and be determined with improving by gps receiver
Geographical coordinate accuracy.GPS module can provide information to PTCC modules.The information provided by GPS module can include:
Time (e.g., UTC, local);Geographical coordinate (e.g., longitude & latitudes, northern line & east line);Control information (e.g., WAAS, difference
Not);Speed;And direct of travel.
DAP can receive (e.g., it is determined that, detection, request) with train, train system (e.g., hardware, software) and/or row
The relevant information of the running status (such as train status) of car.For example, DAP can be received from the system of train with the speed of train,
Train acceleration, train deceleration degree, breaking force (power such as the applied), brake pressure, braking circuit state, Train Wheel are led
The information that gravitation, inertia are measured, fluid (e.g., oil, hydraulic pressure) pressure and energy expenditure are relevant.Can be via being used by train
Signal bus provides the information from train, to convey the information relevant with the state of train system and operation.Signal bus bag
The signal bus of one or more routines is included, such as fieldbus (such as IEC 61158), MVB (" MVB "), is had
Line train bus-line (" WTB "), control area network-bus (" CanBUS "), TCN (" TCN ") (such as IEC 61375)
And Process FieldbusROFIBUS (" Profibus ").Signal bus can perform including the use of any routine and/or proprietary protocol to be had
Line and/or the equipment of wireless (such as TTEthernet) communication.
DAP can also include any conventional sensor, to detect the system of train without the information provided.Sensor can
With any opening position of deployment (e.g., be attached, install) ON TRAINS.Sensor can be directly and/or via another equipment or total
Line (e.g., signal bus, control unit for vehicle, wide train bus-line, MVB) provides information to DAP.Sensor can
To detect any physical property (e.g., density, elasticity, electrical properties, flow, magnetic property, momentum, pressure, temperature, tension force, speed
Rate, viscosity).DAP can provide the information relevant with train via PTCC modules to other modules of the PTC ecosystems.
DAP can be via PTCC modules from the PTC ecosystems any module receive information.DAP can be via PTCC moulds
Block provides the information received from any source to other modules of the PTC ecosystems.Other modules can use by DAP provide or
The respective function of these modules is performed by the information of DAP offers.
DAP can store received data.DAP can access the data of storage.DAP can create received data
Historical record.DAP can make the data from a source related to another source.DAP can make a type of data with
Another type of data are related.DAP can handle and (e.g., format, manipulate, inferring) data.DAP can be stored can be at least
Be partially used for drawing train advance where the signal lamp state of track, the geographical position of train and for active train
The data of the other information of control.
DAP can be from PTCC module receive informations.The information received by DAP from PTCC modules can include:To train shape
The request of state data;Request to brake interface state;Activate the order of train behavior (speed, braking, traction force);It is right
The request of failure message;Confirmation to failure message;The request of alarm is sent ON TRAINS;To the logical of the alarm that is sent on train
The request known;And the request to roadside frastructure state.
DAP can provide information to PTCC modules.The information provided from DAP to PTCC modules can include:From train
Signal bus the data on train status;Confirmation to request;Failure message on train bus-line;And roadside facility
State.
Environment around VA modules detection train.The environment that VA modules detection train travels through.VA modules can detect
Track that train travels over, the track adjacent with the track that train is advanced, roadside (such as along track) signal (arm plate, machine
Tool, lamp, position) aspect (as occur), infrastructure (e.g., bridge, viaduct, tunnel) and/or object (e.g., people, animal,
Vehicle).Other example includes:PTC assets, ETCS assets, track, signal, signal lamp, permanent constraint of velocity, contact net
(catenary, stretched wire) structure, contact net lines, rate limitation direction board, wayside security's structure, crossroad, at crossroad
Pavement, track switch (switch, device for shifting gears) the headroom point position on main orbit and by-track road, headroom/structure gauge/
The beginning and end limitation in no signal region of Dynamic Envelope (envelope), track detecting circuit, shed, website, tunnel
Road, bridge, turnout, inclined-plane, curved surface, track switch, sleeper, railway ballast, culvert, discharge structure, vegetation passage (ingress), railway frog (two
The crosspoint of bar rail), highway grade crossing road, integer milepost, interchange, interlocking/control point position, safeguard set
Apply, mileage direction board and other direction boards and signal.
VA modules can detect ring using the conventional sensors of any kind of detection physical property and/or physical characteristic
Border.The sensor of VA modules can include video camera (e.g., still camera, video camera), distance sensor (such as optical detection
And ranging), radar, infrared ray sensor, motion sensor and range sensor.The operation of VA modules can be dependent on train
Geographical position, track condition, environmental aspect (such as weather), the speed of train.VA operation can be included to collection information
The selection of sensor and the selection of sampling rate to sensor.
VA modules can be from PTCC module receive informations.The information provided by PTCC modules can provide to control VA moulds
The parameter of the operation of block and/or setting.For example, PTCC can provide the sampling frequency of one or more sensors for controlling VA
The information of rate.The information received by VA from PTCC modules can include:The frequency of sampling, the threshold value of sensing data and it is used for
Timing and the sensor configuration of processing.
VA modules can provide information to PTCC modules.The information provided from VA modules to PTCC modules can include:Mesh
The configuration parameter of front sensor, operating condition sensor, sensor capability (e.g., scope, resolution ratio, maximum operational factor), original
Beginning sensing data or sensing data, disposal ability and data format through processing.
Original sensor data or sensing data through processing can include point cloud (e.g., two-dimentional, three-dimensional), an image
(such as jpg), image sequence, video sequence (e.g., live, record playback), the map (e.g., two-dimentional, three-dimensional) of scanning, by light
Detection and image, infrared image and/or the twilight image (such as night vision) of ranging (such as LIDAR) detection.VA modules can be to sensing
Device data perform some processing.Processing can include data reduction, data enhancing, inferred from input data and Object identifying.
Sensing data can be detected and/or identified by VA modules and/or PTCC resume modules:The rail that train uses
Road;To the distance of track, object and/or infrastructure;Wayside signal indicates (e.g., implication, message, instruction, state, situation);
Track condition (e.g., transitable, substandard);Track curvature;The direction (e.g., turn, keep straight on) for the section that will be reached;Partially
From the track (e.g., descending, upward slope) of level;Intersection;Crossroad, interlocking exchange;The train drawn from environmental information
Position;And track identities (such as track ID).
VA modules can couple (as installed) to train.Any opening position that VA modules can couple ON TRAINS (e.g., is pushed up
Portion, inside, bottom).The coupling can be fixed and/or adjustable.The sensor of VA modules is permitted in adjustable coupling
Viewpoint moved relative to train and/or environment.The adjustment of VA position can be carried out manually or automatically.The adjustment can
Made with the environmental aspect and the operating condition sensor that are in response to around the geographical position of train, track condition, train
's.
PTCC is using the access of its all subsystem (such as module) to PTC system come according to the sensing obtained from VA modules
Device data draw (e.g., it is determined that, calculate, infer) track ID and signal lamp state.It is discussed above in addition, PTCC modules can utilize
Train operation state information and data from gps receiver, to refine geographic position data.PTCC modules can also make
With from PTC environment --- including PTC vision systems --- any module information, to describe and/or understand by VA modules
The sensor information of offer.For example, PTCC can use the geographical location information from GPS module to determine the base detected by VA
Whether Infrastructure or signaling data correspond to ad-hoc location.The speed drawn from the video information provided by VA modules and course
(such as orientation) information can be compared with the speed and course information provided by GPS module, to verify accuracy or determine just
The possibility of true property.PTCC can use the image that is provided by VA modules and positional information from GPS module to prepare warp
The cartographic information of operator is supplied to by the user interface of HMI module.PTCC can use current data and history from DAP
Data are so that using the position of dead reckoning detection train, position determines can be with the position by VA modules and/or GPS module offer
Confidence manner of breathing closes.PTCC can be ask from other trains or roadside radio via VCD modules and be answered machine (such as transponder) reception communication
Determined for position, the position determines can by using the positional information from VA modules and/or GPS module or even
Dead reckoning positional information from DAP carries out related and/or correction (as refined).Furthermore, it is possible to operator is asked via HMI
User interface input track ID, signal lamp state or train position, for further related and/or checking.
PTCC modules can also provide information to attendant via HMI user interfaces and (e.g., message, police are called in action
Action, the order show, suggested).Using control algolithm, PTCC can get around attendant, and utilize and brake interface or traction
The speed of the integration adjustment train of interface promotes the change of train behavior (e.g., function, running).PTCC is by describing data
Recipient interested, Payload, frequency, route and the duration of stream handle the route to information, to be listened with third party
Many and collaborative share train status.
PTCC can also automatically assign/packet of receive information, or pass through the shared rear end in control room
Server or from railroad operator or from control room terminal or from attendant or from wayside signal lamp or PTC
Module in vision system is subscribed to the action of third party audience of data on train and called to assign/receive information
Packet.
PTCC can also receive the relevant information of assets near the position with mobile vehicle.PTCC can be adopted using VA
The collection data relevant with PTC and other assets.PTCC can also handle most freshly harvested data (or forwarding the data) to audit
Or the information in enhancing back-end data base.
Algorithm:Determine which rail rolling stock is currently using in the track recognizer (TIA) that Fig. 6 describes into Fig. 7
Road.TIA by by the feature from 3D LIDAR scanners and FLIR video cameras overlap onto on vehicle-mounted vidicon frame buffer come
Create Superposition Characteristics data set.The superset of feature (global characteristics vector) allows three orthogonal measurings of track and three-dimensional composition.
Thermal characteristics from FLIR video cameras can be used for the thermal characteristics that identification (e.g., separates, positioned, isolation) railroad track
Signal (signature), to generate region of interest (space & termporal filters) in global characteristics vector.
The range information of 3D point cloud data set from 3D LIDAR scanners can be used for the height above sea level for identifying railroad track
Highly, with the generation region of interest (space & termporal filters) equally in global characteristics vector.
Wireline inspection algorithm can be used in the 3D point cloud data of vehicle-mounted vidicon, FLIR video cameras and 3D LIDAR scanners
On collection, the confidence level in terms of identifying track with further increase.
Colouring information from vehicle-mounted vidicon and FLIR video cameras can be used for creating sense in global characteristics vector
Region of interest (space & termporal filters).
TIA can search the overlapping of region of interest from the multiple orthogonal measuring to global characteristics vector, to increase track
The redundancy and confidence level of identification data.
When the region of interest in global characteristics vector does not have overlapping, TIA can be filtered using the data of region of interest
Fall false judgment.
TIA can handle the characteristic vector in region of interest to identify the width of track, distance and curvature.
TIA can check the speed that railroad track is assembled towards a bit, further to verify track identification process;In addition,
The inclination of railroad track can be used for filtering out the noise of global characteristics vector data concentration.
TIA can consider before relative deviation post of the train between multiple railroad tracks is identified and characteristic vector
Room and time uniformity.
It can be moved by carrying out multiple repairing weld to gps receiver to obtain directionality course with being created in geographical coordinate
Dynamic time change profile.
It can be obtained by inquiring about the GIS dataset of local cache or the back-end server of remote hosting possible absolute
Track ID list.
Gps receiver lose it is synchronous with gps satellite in the case of, can be counted using mileometer and directionality course
Calculate reckoning skew.
Relative deviation posts and the ginseng of possible absolute orbit ID list of the TIA by train between multiple railroad tracks
Examine information to be compared, to identify train in the absolute orbit ID used.
After TIA obtains absolute orbit ID, global characteristics vector sample can be annotated geographical position and (sat as geographical
Mark) information and track ID.This allows TIA directly to determine orbital position following using global characteristics vector data collection.The machine
Learning method reduces search absolute orbit ID calculating cost.
TIA can also match the global characteristics vector sample from local or back-end data base and space conversion.It is empty
Between the parameter changed can be used for the deviation post that calculates the reference position with being generated according to match query.
In addition, TIA can utilize global characteristics vector so that (e.g., image mosaic, geometry are matched somebody with somebody using various image processing techniques
Standard, image calibration, image co-registration) by the feature of multiple points or the merging features of a single point in space in space
To together.This is produced into verification (collate) global characteristics vector for the multiple points or a single point having in space
The superset of characteristic.
Using the superset of data, TIA can be it is determined that make relative orbit ID deviation post normalizing before absolute orbit ID
Change.This is useful in the case where track is outer in the scope of sighting device (VA).This function is depicted in Figure 10.
TIA is the core component in PTC vision systems, TIA eliminate to obtain the wireless inquiry of position data answer machine,
The demand of beacon or transponder.TIA can also enable network range (network wide) GIS of railroad operator for them
Data set annotates the railroad track of newest structure, and these network range GIS datasets are drawing roadside facility and infrastructure money
It is authoritative in terms of the map of production.
Signal lamp state algorithm (SSA) described in Fig. 8 determines train currently in the signal lamp state of the track used.Should
The purpose of part is to ensure that the expected operation of railroad operator or Model control room or central control room is deferred in the operation of train
Parameter.It can be examined in the distributed environment for having many back-end servers or in the centralized environment for sharing back-end server
The compliance of inertia measurement of the core train along railroad track.For the track ID used by semaphore lamp state and by train
Correlation, the ability that train obtains absolute orbit ID is important.Once establish semaphore lamp state and absolute orbit ID it
Between correlation, examination & verification signal lamp compliance be possible to.The placement of sensor is for effectively determining semaphore lamp state
For be important.Fig. 4 depict wherein 3D LIDAR scanners forward and installed in train roof top one
Individual embodiment.
SSA considers the absolute orbit ID that train uses, to audit the signal lamp compliance of train.Once complete track with
The correlation of semaphore lamp, the signal lamp state can actuating action from the semaphore lamp are called as to train or row
The feedback of car person.
By analyzing the management regulation to wayside signal from railroad operator, railroad track and semaphore lamp state
Correlation be possible.Using documents and materials are managed, the space-time uniformity of semaphore lamp can be with the sky of railroad track
M- time consistency is compared.Scoring mechanism can be used for selecting optimal candidate's semaphore lamp to be used for what train used
Current rail track.
Local or Remote GIS dataset can be inquired about to confirm the geographical position of semaphore lamp.
Local or Remote signal server can be inquired about, to confirm the signal lamp state of semaphore lamp and PTC visions system
The signal lamp state inferred of uniting matches.
Wherein signal lamp state can be used for confirmation PTC vision systems via radio communication for the region of train
Accuracy, and additionally enhancing is supplied to the feedback of machine learning device, and the machine learning device helps to adjust PTC visions
System.
The 3D point cloud data set obtained from PTC vision systems can be used for the structure for analyzing semaphore lamp.If sense is emerging
The expection specification matching that the structure of interesting object limits with the semaphore lamp being such as directed to by supervision department in the rail path, then should
Object of interest can be annotated and be added as the candidate with reference to scoring mechanism above.
The infrared image captured by FLIR video cameras can be used for identification from the light of roadside semaphore lamp transmitting.From
In the case of candidate's semaphore lamp transmitting feux rouges related to the track that train is currently located, action call will be assigned to
Train-installed HMI is for signal lamp compliance.When train does not defer to the arm plate related to the track that train is currently located
During signal lamp, action is called and will be dispatched directly to train-installed brake interface for signal compliance.
Color spectrum in the image captured by PTC vision systems can be divided to calculate center of fiqure (centroid),
The center of fiqure is used for the spot for identifying the green glow of similar signal lamp, feux rouges, gold-tinted or double gold-tinteds.The space coordinates and its spot of center of fiqure
Size can be used for space-time uniformity using the specification verification semaphore lamp from supervision department.
By analyzing interval between the curvature of track, the rail on track and orbital spacing towards one on horizon
The convergence speed of point, the consistent linearity curve of space-time of track can be created.Arm plate letter can be created by analyzing following component
The consistent linearity curve of space-time of signal lamp:Space distance and phase between point in the height of semaphore lamp, space
For train currently in the orientation and distance of the track used.
Back-end server can be inquired about to be informed to train along the train currently in the expection of the railroad track section used
Semaphore lamp state.
Back-end server can be inquired about to be informed to train along the iron by absolute orbit ID and geographical position coordinates identification
The expection semaphore lamp state of road orbital segment.
The position thinning algorithm described in Fig. 3 provides the geographical positioning service of train-installed high confidence level.The algorithm
Purpose be to ensure that the loss that geographical positioning service will not occur when single sensor fault.PRA is geographical dependent on redundancy
Location-based service obtains orbital position.
GPS or differential GPS can be used for obtaining fairly accurate geographical position coordinates.
Rotational speed counting evidence can be used for calculating deviation post together with directionality course information.
WiFi antennas can scan signal intensities of the SSID together with each SSID when GPS works, and later using Jie
Matter access control (MAC) address (or any unique identifiers associated with SSID) quickly determines geographical position coordinates.By
The signal intensity that WiFi antennas are scanned the SSID of period can be used for calculating the position relative to measurement origin.PTC visions
System can select the confidence level in the geographical position based on current train come by SSID profiles (SSID name, MAC Address, geography
Position coordinates, signal intensity) it is used as in reference point write into Databasce.
The global characteristics vector created by PTC vision systems can be used for searching geographical position coordinates, to further ensure that
The accuracy of geographical position coordinates.
Will be filtered out from the scoring mechanism of above-mentioned all part extraction samples, which may suppress train, obtains geographical position
The inconsistent sample of the ability of information.In addition, performance and accuracy based on each subassembly in PRA, sample can bands
There is different weights.
The high-level process description of PTC vision systems
The flow chart shown in this part, our reference pictures 9.Each subsystem of PTC vision systems from the description above
In train status is sampled.Train status is defined as the Summary of track, signal lamp and on-vehicle information.Especially,
The state by track ID, the signal lamp state of coherent signal lamp, related on-vehicle information, positional information (it is above-described it is preposition and
Rearmounted thinning algorithm, with reference to PRA, TIA and SSA algorithm) and the information composition that obtains from back-end server.These back-end services
Device preserves the information relevant with railway infrastructure.The back-end data base of assets is by mobile vehicle and railroad operator and office worker
Remote access.Mobile train and its attendant for example use the information to expect the signal lamp along route.Operator and Wei
Such as orbit information can be accessed by repairing office worker.These reports and notice and signal lamp and direction board, structure, track characteristic and money
Production, security information are relevant.
After the state is gathered, PTC vision systems make an announcement (Local or Remote), may send police ON TRAINS
Report, and may be automatic by being docked with various onboard subsystems (e.g., drawing interface, brake interface, traction sliding system)
The inertia measurement of ground control train.
The sensing stage
Vehicle-mounted data:Vehicle-mounted data part represents a unit, at the unit, from all numbers of various train systems extraction
According to being collected and can use.The data typically include, but not limited to:Temporal information, the diagnosis letter from various mobile units
Breath, energy monitoring information, brake interface information, positional information, the signaling status for going to roadside facility obtained from train interface,
The ambient condition obtained by VA equipment on vehicle-mounted or other trains and any other contributing to from part are active
The data of Train Control.
The data can be used for miscellaneous part in PTC vision systems, and can be transferred to remote server, other row
Car or roadside facility.
Position data operates in for ensuring train in the safety enveloping line of PTC standards for meeting FRA
It is important in strategy.In this regard, roadside facility is currently utilized to accurately determine vehicle location in industry.Retouch above
The output (e.g., TIA&SSA) for the location-based service stated is based on computer vision algorithms make and provides relative orbit position.
Relative position can be obtained by using single sensor or multiple sensors.The position that we obtain is as skew
Position is returned, and is typically denoted as relative orbit numbering.Directionality course can also establish inquiry with from anti-to train
Feedback obtains the factor of absolute position.
Can from the local data sets of the local data base of caching or caching, remote data base, teledata collection, use car
Relative deviation post, GPS samples, Wi-Fi SSID and its corresponding signal intensity of inertia metric data are carried, or by with showing
Some wayside signal facilities synchronously obtain absolute position.
Various types of data sets that we use include but is not limited to:3D point cloud data set, FLIR images, from vehicle-mounted
The video buffer data of video camera.
Once being aware of position, the information may be used for the signal lamp state from wayside signal and corresponding track
It is related.Location-based service can also be exposed to third party audience.The on-vehicle parts limited in PTC vision systems can serve as position
The audience of service.In addition, train can scan the MAC ID of the networked devices in peripheral region, and any application can be directed to
Filtered using MAC ID in these networked devices used.This is useful, the environment sensing for creating environment sensing application
Using depending on third party device (e.g., cell phone, laptop computer, notebook computer, server in station and other calculating
Equipment) MAC ID and train geographical position pairing.
For ensuring that train always complies with PTC safety enveloping lines, track signal lamp state is important.PTC vision systems
Envelop of function include from wayside signal (semaphore lamp state) infer signal lamp value.In this regard, communication module or
Sighting device can identify the signal lamp value of roadside facility.In the sightless region of signal lamp, center back end server can be with
Feedack is used as dependent on to train.When roadside facility is equipped with radio communications set, the information can also increase
The signal of strong view-based access control model infers algorithm (e.g., TIA&SSA).Used when PTC vision systems are distinguished (discretion)
Data set.
Using the data set gathered by PTC vision systems, people can identify track according to the remainder data in device
Feature, and relative orbit position can be identified.After relative orbit position can be sent to together with directionality course information
Server is held to obtain absolute orbit ID.Absolute orbit ID is represented as the track identities listed by operator.It is this effectively negative
Lotus is arbitrary for train, it is allowed to more seamless operation between multiple operators, without having operator special ON TRAINS
Determine software stack.Operator's independent software allows train to be run with great interoperability, even if train is grasped according to different rails
It is also such that infrastructure is advanced through as member.Because Payload is arbitrary, train is substantially interoperable, even in
It is also such when switching between rail operator.When rolling stock is advanced along track, sighting device is generated for more new assets
Data needed for information.Then the data are processed to verify the integrality of a certain assets information, and update other assets letter
Breath.Then it can detect and report the assets, the assets damaged or the assets having been tampered with of missing.Whenever with sighting device
Vehicle along track travel downwardly when, the state of infrastructure can also be verified, and can assess safe for operation.For example, hold
Row Clearance survey, to ensure the path of no barrier obstruction train.With time Estimate and monitor the body of the railway ballast of supporting track
Product.
Rear end:
Back-end component has many purposes.For one of which, back-end component receives, annotation, number of the storage from train
According to and algorithm, and such data and algorithm are transmitted to the subscriber of various Local or Remotes.The many mistakes of rear end also trustship
Journey, for analyze data (in real time or offline), then generation correctly output.Then the output is sent directly to as feedback
Train, or it is transferred to commander transfer center or train website.
Some in aforementioned process can include:To reduce the spacing distance between train to optimize on some passages
The algorithm of flow;Optimize the algorithm of whole network flow by considering each train or passage;And continue to monitor train
Position and the anti-collision algorithm of behavior.
Rear end also trustship is inquired about by running train to obtain the asset database of assets and infrastructure information, above- mentioned information
As required as rolling stock movement rule.The database preserves the following assets with relevant information and feature:PTC assets,
ETCS assets, track, signal, signal lamp, permanent constraint of velocity, contact web frame, contact net lines, rate limitation direction board, road
Other safeguard construction, crossroad, the pavement at crossroad, the headroom point position of track switch on main orbit and by-track road
Put, the beginning and end constraint in no signal region of headroom/structure gauge/sports bag winding thread, track detecting circuit, shed,
Website, tunnel, bridge, turnout, inclined-plane, curved surface, track switch, sleeper, railway ballast, culvert, discharge structure, vegetation passage, railway frog (two
The crosspoint of rail), highway grade crossing road, integer milepost, interchange, interlocking/control point position, safeguard set
Apply, mileage direction board and other direction boards and signal.
Rolling stock refines track recognizer, position thinning algorithm and letter using the information inquired from database
Signal lamp state detection algorithm.Along train (or any other car using machine vision device for track/move close to track
) gather to insert, verify and update the data the data in storehouse needed for information.Back-end infrastructure is also given birth to for each Railroad personnel
Into the alarm and report relevant with the state of assets.
Feedback stage
Automatically control:
If in the presence of the drying method (application in such as Fig. 5) that can use PTC vision systems control train.The sensing stage it is defeated
Going out can be independently of other any some action of system trigger.For example, detecting that brake interface can be automatic when making a dash across the red light
Trigger to attempt to stop train on ground.
Some control commands can also reach train by the VCD of train.As such, back-end system can for example indicate train
Increase its speed, so as to reduce the spacing distance between train.Other train subsystems can also be activated by PTC vision systems
System, as long as they can be accessed in itself in locomotive head.
Vehicle-mounted alarm:
Feedback can also reach locomotive head and attendant by alarm., can be on HMI in the case where for example making a dash across the red light
Show alarm.The alarm can automatically control or individualism with any.Can be by confirming or independently halting to stop
Only alarm.
Notify (local/remote):
Feedback can reach attendant in the form of notice by the user interface of HMI module.These notices can describe
Pass through PTC vision systems local sensing and the data gathered or the data obtained by VCD from back-end system.These notices can
Audience can be needed, or can be for good and all activated.The example of notice can be on the speed suggestion to be observed of attendant.
Rear end framework and data processing
Rear end can have two kinds of modules:Data aggregate and data processing.Data aggregate is that a kind of effect is intended to polymerization letter
Breath and between train and center back end routing iinformation module.Data processor is used to advise to train.Communication
It is two-way, and the back-end server can service all various possible applications from PTC vision systems.
Possible application for PTC vision systems includes following:Signal detection;Track detecting;Speed sync;Infer rail
The Interlock Status in road and the state is transferred back to other trains in network;Fuel optimization;Collision avoidance system;Track detecting is calculated
Method;Rail deformation detection or preventative derailing detection;Track performance metric;To comprehensive using being created from the sample repeatedly operated
Close the merging algorithm for images with reference to data set;Believe for such as preventive maintenance, fault detect and/or by the vibration performance of train
Number intersection train imaging;Geographical position or geographical filtering services based on imaging;Geographical position or geographical mistake based on SSID
Filter;And GPS+ inertia measurement+sensing fusion based on computer vision algorithms make.
According to other embodiments, remote sense and location feature can be used for realizing automotive vehicle such as autonomous driving
The run time system of automobile.Figure 25 is the schematic block diagram of the exemplary on board formula system for vehicle location and/or control.
Vehicular run time engine (" IVRE ") 2500 and Vehicle Decision Method engine 2510 are being typically based on of locally being realized on vehicle
The calculating of microprocessor and control module.Local 3D cache maps device 2530 stores and in vehicle such as by GPS and IMU sensors
The associated map datum in region around 2520 rough positions determined, and can (it can be with via communication module 2540
Including such as cellular data transceiver) regularly or continuously it is updated from long-range map storage device.Machine vision sensor
2550 can include one or more mechanisms of the home environment for sensing du vehicule, such as LiDAR, video camera
And/or radar.
In operation, IVRE 2500 from vehicle GPS and IMU sensors 2520 by obtaining rough vehicle location come real
Existing vehicle location.The environmental characteristic signal of the home environment of the generation instruction vehicle periphery of vehicle vision sensor 2550, the environment
Characteristic signal is passed to IVRE 2500.IVRE 2500 uses the environmental characteristic signal received from machine vision sensor 2550
Local 3D cache maps device 2530 is inquired about, by the feature or object observed in the home environment of vehicle and is stored in caching
Known features with known location or object in 3D semanteme maps in device 2530 are matched.By by the relative of vehicle
Compared with the position on map of position and those features and object for observing of local feature or object, can with
Usually using the possible positions that vehicle is refined compared to notable more accurate mode of GPS, wherein error span can with centimetre
Measurement.
Detailed vehicle location and other information observed or calculated can be used for realizing other functions, such as vehicle control
System and/or map examination & verification.It is, for example, possible to use figure as described elsewhere herein is analyzed with other data analysis mechanisms
Data from machine vision sensor 2550, for making IVRE 2500 determine the center line in the track where vehicle traveling.
IVRE 2500 can also be run to obtain the semanteme (such as event and triggering) along the route of vehicle.Available computing resource can
For by by from the map of concentration obtain before the assets information observed (and be such as stored in local 3D cache maps
Assets information in device 2530) carried out with the assets information drawn from the real time data captured by machine vision sensor 2550
Compare the map datum resource come in audit set.Thus IVRE 2500 can identify error of omission (that is, from the map number of concentration
According to the observed assets of middle omission) and fault (commission) mistake (that is, not by machine in the map datum of concentration
The assets that device vision sensor 2550 is observed).Such mistake can be stored in buffer 2530, and then via logical
Letter module 2540 is sent to center map repository.
In some embodiments, examination & verification of the local vehicle to map datum can be by centralized Control startup of server, should
Centralized Control server is via communication module 2540 and vehicle communication.If for example, since the last time examination & verification map section is gone over
Time exceed threshold value, centralized Control server can ask the examination & verification from the local vehicle for being advanced through target area.
In another embodiment, if a vehicle report concentration map datum and locally observe situation between difference, that
Other one or more vehicles that centralized Control server can ask to move in the region with the difference confirm examination & verification.
Examination & verification request may relate to the various combinations of geographic zone and/or mapping layer.
In some embodiments, it may be desirable to using information such as accurate vehicle location, assets and semanteme and lead
Information of navigating is as the input to Vehicle Decision Method engine 2510.Vehicle Decision Method engine 2510 can be run to control various other systems
With the function of vehicle.For example, in autonomous driving implementation, Vehicle Decision Method module 2510 can utilize lane center information
With accurate vehicle position information, to manipulate vehicle and maintenance centre's lane position.Use system described herein and mistake
Journey can be advantageously carried out the vehicle control system of these and other.
Semantic map is created using geographical spatial data
Map is the set of object, the position of object and Properties of Objects.Map can be divided stratification, wherein every layer equal
It is the group of the object of same type.The position of each object be defined together with geometric attribute (such as:The position of electric pole
Can be the point in three dimensions, and signal lamp can be positioned by drawing polygon around electric pole).When also recording
During semantic association between different objects and layer, map becomes " semantic ".Various direction boards and center are have recorded for example, working as
During association between line, the ground that is made up of the center line in various tracks on railway and the direction board around infrastructure
Figure is denoted as semanteme.This can be marked by creating the unique identifiers of direction board and the unique of the track relevant with direction board
The mapping between symbol is known to realize.The semanteme of map turns to the vehicle of area of consumption figure or user creates more linguistic context.Semantic map
It can also be packaged together with the management information from each shipping office.
The physical geometry of any assets described in map.For describe shape geometric properties include point,
Line, polygon and camber line.Feature generally three-dimensional, but they can project to the two dimension that wherein depth/height above sea level is lost
In space.Generally, it can record and transmit semantic map with different coordinates and reference system.There is also allow from a seat
The map projection of reference system is marked to the conversion of next coordinate reference system.Can be packed these maps, and in a different format
Transmit these maps.General format is including GeoJSON, KML, shapefile etc..
In some embodiments, imaged for creating the geographical spatial data of semantic map from LiDAR, visible spectrum
Machine, thermal camera and other optical devices.The action for obtaining the machine vision data for creating map is referred to as surveying,
Wherein machine vision data is geographically with reference to tellurian ad-hoc location.Output is one group of three-dimensional data points together with visible spectrum
With the image and video feed of other frequencies.There may be many different hardware platforms for data acquisition.Collection vehicle
And variable (antenna, mobile, earth station).First, in the case where collection vehicle is located at the origin of reference system
Gather geographical spatial data.Pass through positioning vehicle (use such as Inertial Measurement Unit (IMU) and global positioning system in whole exploration
Unite (GPS)), then image, laser scanning and video feed are registered to the fixed reference frame as geographical reference.In exploration
The data of generation can be streamed (stream, stream transmission) or be saved locally for consumption later.
Some embodiments of vehicle location described herein and home environment sensing system have benefited from the exploration of point of use cloud
Data.The current or prospective home environment that the semantic map drawn from a cloud survey data can provide with vehicle for vehicle has
The high-level details and information of pass, it can be used for such as auxiliary phase and does decision system to vehicle location or as autonomous control
The input data of (e.g., Braking mode, manipulation, speed control etc.).Additionally or alternatively, the cloud data measured by vehicle can
So that compared with the cloud data previously measured, to detect the situation of home environment or change, that such as falls down sets, is excessively raw
Long vegetation, the sign board of change, track block, track or road barrier etc..Detected environmental change can be used for
Further update semantics map.
However, the possible roughly the same big data set of level of increase point cloud survey data details, this is provided for service
All it is probably high cost or time-consuming for person's processing or for vehicle storage or processing.For example, along rail track
The 3D railway reconnaissances system based on LiDAR that straight line is advanced can be geographical space number of the scanning generation more than 20GB per km
According to.Then the original point cloud data that generation is scanned by LiDAR usually requires that extra processing to extract useful assets information.
Conventionally by using 3D vision softwares three-dimensional semantic map is created from cloud data and other geographical spatial datas.
Figure 11 A show the typical prior art process for the withdrawal of assets information from cloud data.In step S1100, survey
Program Generating cloud data collection is surveyed, such as using LiDAR surveying devices.In step S1105, original point cloud data is visual
Change.Generally, GIS-Geographic Information System (GIS) analyst is come in manual identification, annotation and grouped data using sensing-click method
Key Asset.The first step during GIS analyses, which is the terabyte of cloud data being separated into, smaller manages section.This is
Because current personal computers are limited (memory/computation ability) and can not manage the LiDAR data of terabyte at once
The fact.
Then, GIS analysts travel through each in more dot cloud section using 3D visual softwares.Because they advance
By their corresponding sections, GIS analysts are described and annotated to important assets.Finally, each GIS analysts through note
The assets released are incorporated into a map (step S110).The number of separation may merged by changing file format and software systems
Extra difficulty is produced according to concentrating.
Both prior art process and infrastructure are limited to from cloud data extraction of values.Sensing-click annotation is manual
, slowly and be easy to malfunction.In addition, traditional system based on file prevents GIS developer and manager from effectively managing
Manage the cloud data collection increased.
Figure 11 B show the alternative from original point cloud data withdrawal of assets.In step S1150, surveyed
To generate original point cloud data.In step S1155, assets map is directly generated according to original point cloud data, without requiring pair
Substantial amounts of complex data collection is visualized, or does not require the manual annotations data.
Figure 12 shows the computing device for concentrating quickly and efficiently withdrawal of assets information from big cloud data.Figure
13 show the process for the device using Figure 12.Preferably, using the connection internet including one or more servers
Cloud computing resources realize the part in Figure 12 device.Front end component 1200 includes data uploading tools 1205, configuration work
Tool 1210 and map retrieval instrument 1215.Front end component 1200 is provided based on interacting and controlling with computing device for terminal user
Calculate the mechanism of device.
Using data uploading tools 1205, user can be by LiDAR and other explorations from local data storage device
Data are uploaded to data storage part 1220 (step S1300).Data storage part 1220 can be realized for data storage
Distributed file system (such as Hadoop distributed file systems) or other mechanisms.Configuration tool 1210 can be via user's
The computing device (not shown) of networking accesses, and can allow the user to limit the data uploaded and other exploration details
Form, and specify the assets (step S1305) searched and annotated.Interacted in user with configuration tool 1210 to select desired money
After production, various options have been provided the user to configure output map form.Preferably, then configuration tool 1210 is used from configuration
Family is solicited the desired turnaround time, and the estimated cost (step S1310) for analysis is presented for user.Cost estimate is base
In the size of upload data set for example to be analyzed, the quantity (and complexity) of the assets of selection, output format and selection
What the turnaround time determined.Finally, when configuring completion, user is interacted with configuration tool 1210 to start analysis work (step
S1315)。
The geographical spatial data of the upload of front end 1200 is traced back through in database collection.The data pass through classification, geography
Area and other properties carry out tissue.As data were developed by the various execution stages, Relational database entry is updated.
Stored the cloud data that front end tool uploads by safety and in a manner of repeating.In order to simplify retrieval, data
Different size of segment is paved into cartesian coordinate system.Therefore segment is limited in two dimensions and correspondingly existed in itself
In NameSpace.Preferably, segment is limited to X and Y sizes, and is not only restricted to the vertical or centrifugal force direction parallel to the earth
Z sizes so that segment defines highly unrestricted (that is, the scope for being limited solely by obtainable geographical spatial data) and had
There is the cylindrical region of square-section.In exemplary implementation, the figure that side (along the horizontal plane) is 1000m can be utilized
Block.Represent then the file of segment belongs to preservation the institute of the specific geographical area limited by the segment a little, without preserving it
He point.In certain embodiments, tree construction (such as quaternary tree and Octree) is realized according to the traversal mode of data.
The data processing that semantic map is automatically extracted from geographical spatial data is the calculating realized in processing unit 1240
(embodiment that computing cluster is further described below with reference to Figure 16) occurs on cluster.These have passed through network-accessible
Memory cell 1220 have accessed a cloud and other data.Similarly store intermediate result and final result.
Figure 14 shows the process that can be performed when analyzing work and starting by Figure 12 device.In order to simplify at data
Reason, and make it possible to realize that map reduces data analysis frame work, cloud data is by data storage/pretreatment component 1200
Segment blocking (step S1400).These blocks can be subset or the combination of segment of segment, may be chosen to optimize such as phase
The processing method of prestige, available memory and other run times consider.It is each (i.e. in processing unit 1240) in computing cluster
Then individual node can be handled combines associated geographical spatial data with the segment subset of the i.e. selection of data-oriented block or segment
With other data.
The density of point cloud is to determine the quantity (or size of segment subset) that handle segment in same calculate node
Key factor.In an exemplary, Figure 15 shows the segment of the quantity relative to the point in exemplary data sets
Size (being represented by diagonal), and the distribution of the segment size of exemplary data sets, the exemplary data sets include along
The LiDAR point cloud data of 2km railway divisions measurement (each segment across cornerwise hachure by representing).Data storage and pretreatment
Part 1220 performs segment polymerization and/or subdivision before data are fed into processing unit 1240, to optimize analytical performance.
The benefit polymerizeing in view of segment as described above, when there is the point cloud density reduced can cause reduced processing
Between.However, low-density would generally make feature detection process more difficult, and higher rate of false alarm can be caused.Cloud data
Abundanter, detection process just becomes more accurate.
Once startup is handled, job scheduler 1225 is created that the queue for including the task relevant with operation, such as in step
Configured in S1305 and S1310.Job scheduler 1225 (generally realizes one or more of analysis institution 1250 each
The different data analysis algorithm of kind) (step S1405) associated with task, and clusters of machines is created in processing unit 1240
Carry out processing data (step S1410).In view of individual node gathering to the segment with known mean size (such as 250MB)
Close the average time of the previous measurement of the data analysis mechanism 1250 of implementation demand, it may be determined that the size of cluster (that is, calculates section
The quantity of point) to meet turnaround time for being asked in step S1310.For example, it is contemplated that it is submitted the sample data for processing
Collection, estimates to spend the calculating time of about 240 hours on eight core desktop computers.Due to data analysis mechanism 1250 preferably by
It is designed to run simultaneously, so job scheduler 1225 can start every cluster with 20 machines of four kernels, and
And identical data set was alternatively handled in approximate 24 hours.
Processing unit 1240 is made up of the set of computing cluster.The size of cluster depends on the quantity of operation.Figure 16 is shown
Example calculation cluster.Each cluster includes:Master instance 1605, is responsible for cluster;The main calculate node of quantity is set
1610, it is also stored data in data-storage system 1220;And " spot " example 1620 of variable number.In some realities
Apply in scheme, it would be desirable to which the size of principle example 1610 is set can handle whole data and meet the turnaround time
It is required that wherein spot example 1620 is activated based on such as their cost and/or activity duration constraint.In other embodiments
In, the computing cluster being made up of completely by spot example or completely main node can be utilized.
Once generating appropriately configured computing cluster, data storage and preprocessor part 1220 are just by the stream of data block
(polymerization for e.g., meeting the segment of desired data subset size) is directed to processing unit 1240 (step S1415).Processing unit
Main node and spot example in 1240 perform appropriate data analysis mechanism 1250, are provided with for example being extracted from 3D point cloud segment
Production or characteristic information.
Once data set has been processed, unit 1240 handles and extracts desired information, will trigger map generator
1230.The output of node in processing unit 1240 is combined into semantic map (step S1420) by map generator 1230.Pass through
Operation inquiry can draw report analysis to analyze specific assets and combinations thereof according to semantic map.
Map generator 1230 can also include annotation integrity verification device, and the annotation integrity verification device is run to verify
The data set of annotation with the time integrality.In some applications, survey location can be repeated in different time.For example, in iron
In the application of road, periodically the railway of equal length can be surveyed equipped with LiDAR train or other railway reconnaissance vehicles,
Such as monitoring health or situation along the assets of track.In the application of some roads, equipped with LiDAR exploration vehicle
It can be advanced in given part of the different time along road.Other roads application in, can with periodic analysis by equipped with
The data of LiDAR motor vehicle such as autonomous driving automobile capture, it is frequent so as to provide the possibility of the home environment of given position
Analysis., can be by assets or local characteristic information when each map generator 1230 generates the new map on given area
Compared with these information included in older map.When a discrepancy is detected, alarm, notice or thing can be triggered
Part.
The output of map generator 1230 finally can be used for user's (step via front end 1200 and map retrieval instrument 1215
S1425).Once operation is completed and generates map, scheduler 1225 (monitoring task and the state of operation) will be terminal
User generates notice.
Characteristics map (position, geometry and feature only comprising various assets) and semantic map can also be stored in
In geographical data bank capable of making remote access.Map datum can be retrieved directly or by server, in order to inquire about and collect knot
Fruit.Can be with complete search map or by selecting specific area-of-interest to retrieve map.
Security, compression and integrality
Data and the security of map can be the importances of many embodiments.Preferably, data upload step
S1300 is using the End to End Encryption (such as AES encryption) from user data source to cloud computing platform.This encryption can be also used for
Communication between the system of user and front end 1200.
In some embodiments, it may be desirable to original point cloud data is stored in data storage part in the compressed format
In 1220.For example, there is the example distributed meter of the storage of a terabyte for every four CPU (CPU) kernel
Cluster is calculated, slower processing time may be caused with its primitive form storage 3D point cloud data, because storage infrastructure
To be that I/O is constrained, while CPU core is sometimes what is left unused.This means before processing data, CPU will be essentially waited for
From memory read data.Being compressed permission system to it before original point cloud data is stored spends less time to read number
According to write data into disk.Therefore, data storage part 1220 can include compression mechanism, with compression point before storing
Cloud data.
However, the original point cloud data by storing compression, processing time can increase, because being analyzed in application data
Data must be decompressed by decompression mechanism before mechanism 1250.Generally, the compression ratio of compressed data is conciliate with compression
Positive relationship be present between the amount of processing time needed for compressed data.Therefore, may expect in some embodiments continuously
Measure CPU time and modulation data compression ratio, with make as closely as possible from memory unit 1220 read data speed and
The rate equation that data can be decompressed and handled by processing unit 1240.
As described herein, many lossless data compression mechanisms can be used for handling a large amount of cloud data collection.Example bag
LempelZivOberhumer (LZO) is included, GZIP (also based on LempelZiv methods) and LASzip are (by rapidlasso GmbH
Issue, and hereinafter referred to as LAZ).Figure 17, Figure 18 and Figure 19 show the comparative analysis of these three compression mechanisms.In compression side
Face, LAZ methods show the constant CPU time on all compression levels, and (compression level is higher, and compressed output file is got over
It is small).This method haves a great attraction, because when compared with LZO and GZIP, this method produces smaller file
Size.However, LZO and GZIP is optimization for decompression, and therefore it is rendered as in terms of the CPU time needed for decompression
To LAZ excellent replacement.In some embodiments, it may be desirable to pass through the essence based on data set and available calculating basis
The characteristic (such as cost and availability) of facility selects compression mechanism to accelerate data from multiple mechanisms with different qualities
Processing minimizes storage demand simultaneously.
Machine vision analysis institution
It is typically based on and it is expected to select data analysis mechanism 1250 from the essence of the information of cloud data extraction.It may be desirable to
Be design with low-down rate of false alarm while keep the mechanism 1250 of acceptable verification and measurement ratio.In order to increase generated map
Confidence level, in some applications, can manually verify knot by checking initial cloud data and raw image data
The subset of fruit.
Track detecting and traversal
In the embodiment of processing railway point cloud survey data, track detecting is probably the important first step.Track is examined
Survey be probably it is important because the confirmation of orbital position contribute to identify assets, and because rule be usually and rail
The relevant each asset allocation ad-hoc location in road.
Figure 20 show can by the step S1415 such as Figure 14 that processing unit 1240 is implemented be used for track detecting and
The process of traversal.In step S2000, identify the cloud data of 100m × 100m sections for analysis.In step S2010,
Analyze 10000m2The geometry of point cloud section, to extract the subset of the point associated with track.Can be using many technologies come real
Existing desired result.In some embodiments, the track of the previous class from similar data sets can be studied to identify rail
The property of data near road, wherein these properties are used as the mark of orbital position in the data newly analyzed.Other technologies bag
Include:Point is projected in two-dimensional space (based on such as height or pulse strength), and using rim detection mechanism and
Conversion belongs to the region of track to separate.Under exemplary service condition, the 10000m in step S20002Point cloud section can be with
It is made up of about 1GB data, and the track subset extracted exported in step S2010 can be made up of about 1MB data.
Figure 21 is enter into step S2000 10000m2Point cloud section and the rail extracted that is exported in step S2010
The visualization of track data.Line 2100 represents the visible track in a cloud.Line 2110 represents the quilt in LiDAR data gatherer process
The track covered, there is the position of estimation.This is typically the result of shadowing effect, and the shadowing effect is when object of interest is hidden
Measuring instrument directly beyond the invisible when the process that can occur.Round dot 2120 corresponds to the problematic of LiDAR tripod systems
Positioning, this causes some orbital segments to be blanked.By using the known spatial continuity property of infrastructure (such as relative to it
The interval for the element that he observes) position (step S2020) of invisible track can be speculated.
Geographical spatial data shows utilizable many dimensions during assets are extracted.It can be sensed with combination image
Device, infrared sensor, video feed and/or multispectral sensor detect confidence level and accuracy to increase.Most of LiDAR systems
System includes the ionization meter for each putting.By analyze on track and track outside point intensity, can by sorting mechanism and
Wave filter is added in system, for improving track detecting rate.Figure 22 is the point cloud in exemplary track detection implementation
The histogram of intensity rank.Figure 22 a show the intensity rank each measured in the analysis body as the cloud data of entirety
Amount.Figure 22 b show for be identified as in a cloud with track corresponding to the identical histogram put.In some cases, simply
Bandpass filter can effectively further constriction belong to track point search space.Other classification sides can also be utilized
Method.
Figure 23 be a part for the output for the implementation for including track detecting mechanism and other assets testing agency can
Depending on change.Via the operation of track detecting mechanism, orbital segment 2300 is identified first, then establishes centre line marks for each track
2310.Once identifying track and track centerline, subsequent analysis component can travel through the track in cloud data, enclose simultaneously
Each point in center line enjoys 360 degree of views of high-resolution cloud data.
Other analysis institutions identify and positioned the other assets or feature being included in semantic map.For example, overhead line is examined
Survey mechanism to identify and position overhead line, and overhead line is demarcated with overhead line centre line marks 2320.Electric pole testing agency identifies
And positioning track by electric pole, and with the positioning track of mark 2330 side electric pole.These and other features can be included in through
By in the semantic map output of system and method described herein generation.
In some embodiments, sequentially applied analysis mechanism, the output of one of mechanism another machine can be used as
The input of structure.For example, in railway applications, the assets and element regularly change, added, remove or shift home environment.Can
The headroom of track above and around can be desirably inspected periodically, to ensure safe operation, and ensures that railway car will not be with
Any bar contact.In this applications, all track detecting mechanisms described above may be implemented as a series of analysis institutions
A part.The output of track detecting mechanism including track centerline can be subsequently used as the defeated of track headroom inspection body
Enter.Relative to track centerline limited boundary frame, and report any object occupied in the border.Bounding box can be changed
Size is with suitable various standards.
The position of direction board, signal lamp, track switch, roadside unit etc. can also be determined using detection framework.Once positioning, root
According to the specification of manufacturer or other object definitions, can be given in the case where providing the geometric properties of each assets to these moneys
The classification of production.
Can valuably adopted another analysis institution be overhead transmission line inspection in railway applications.Can be in a cloud
Identification overhead line in data.Assess height of the electric wire compared with track.Report has the region of sagging circuit.By using electricity
Line bar positional information, the contact mesh-shaped of electric wire can also be assessed.
Although railroad track detection context in describe some analysis components, it is contemplated that and understand, can
So that other kinds of assets are potentially identified using similar analysis institution and method in other application.For example, it is similar to
The mechanism of track detecting mechanism described herein is probably to have in the road context for identifying lane markings and/or curb
.
Computation paradigm
Automatically extracting for map can be by being combined into directed acyclic figure (hereinafter referred to as " figure ") come real by calculation block
It is existing.Block included in these figures has different complexities, from it is simple be averaging and thresholding to conversion, filtering,
Decompose etc..The output in one stage of figure can be fed to any other follow-up phase.These stages need not transport in order
OK, can run parallel, as long as each stage is presented enough information.When creating characteristics map, figure is usual
For classifying to belonging to same category of point in a cloud, or for vectorization.Vectorization refers to by defining line or more
One group of point of the center of side shape, border, position etc. creates (typically imaginary) line or polygon.Can be with as such, calculating figure
For realizing grader, clustering method, fitting routine, neutral net etc..Rotation and projection also often handle skill with machine vision
Art is used in combination.
In order to make full use of Distributed Calculation, creating semantic map according to geographical spatial data can be carried out parallel.In the presence of
The parallelization rank that can much realize.In highest level, survey data can be divided into the region of interest of conventional shape, this
A little region of interest are streamed to different machines and CPU processes.Once all processes are completed, the result from each region just needs
To be merged in " reduction " step, similar to Figure 14 process.Due to the appearance of boundary condition, therefore used in processing most
Those lopsided parts of adjacent edges would generally be removed by filling up region of interest except the excessive data being truncated afterwards.Region of interest it is big
It is small and fill up thickness and determined by the figure of withdrawal of assets or feature.
In another rank, when being handled along the vector of preextraction, parallelization can occur.For example, when in iron
, can be by the region that is extracted in along the track centerline previously extracted around path point when direction board is nearby searched in rail road
Carry out ergodic data.Then multiple processes can be used parallel along the different path points of track.
Finally, when analyzing specific region, each point can individually be considered.In this traversal method, generally extraction
And analyze the volume elements around the point.Those situations of the operating result of any other point are not influenceed in the operating result of a point
Under, the process can also be parallel.
These are some traversal methods used during map building, and some of which method can and be advanced
Row data processing.In addition, GPU (graphics processing unit) also brings very big speed improvement with being used in combination for conventional CPU, and
It can further help to reduce the turnaround time.
Geographical spatial data is not limited to a cloud, but expands to image, video feed, multispectral data, RADAR etc..For
Increase map accuracy and correctness, some embodiments can utilize available any additional data source.It can utilize several
Kind of technology is used from not homologous data.In some embodiments, before data set is fed in calculating figure,
Data set can be combined at pretreatment stage (e.g., step S1400).It this method provide to utilize and be used to locate from multiple sources
The calculating figure of the data of reason.In other embodiments, can be generated using one group of data on assets and its property
Assuming that;Then the data from other sources can be used via other analysis institutions to verify and/or strengthen hypothesis.
Machine learning:
Many machine learning techniques can be realized to help semantic map building process.Existing annotation map can be used for
Training figure simultaneously optimizes to it, to automatically generate accurate semantic map from geographical spatial data.It is input to machine learning
The data of system are made up of survey data and corresponding annotation output map.The output of machine learning system is a refinement figure
Shape, then it can apply to wider survey data, to extract map on a large scale.In some cases, classification
Point cloud is used to be fed in training process (based on the assets belonging to it to each point distribution classification).In other respects, to
Systems of Quantitative Map is used for Learning Map and creates process and adjust processing figure.Dependent on assessment performance (passing through error measure) and it is expected
The reinforcement of performance, these methods are included into supervised learning classification.
The embodiment that Figure 24 shows the system for implementing supervision machine study, including training component 2400 and map generation
Part 2410.Training component 2400 receives original point cloud data 2420 and sample output 2422 as input.In some cases,
Sample output 2422 can be approximate with total data set 1% associated empirical tests output data.Sample output 2422 can be with
Cloud data (point for wherein belonging to special assets classification is grouped together) and/or vectorization map including classification (have
Point, line and the polygon drawn in asset of interest).Training component output 2424 defines the sorting mechanism of optimization, such as
The algorithm coefficient of the analysis institution suitable with the mechanism 1250 in Figure 12 map generation system.Training component output 2424 may be used also
To limit region of interest so that algorithm is most effective, the functional block calculated in figure that be utilized is limited, and/or limit and examined
The feature of interest of the special assets of worry.Training component output 2424 together with whole corpus of original point cloud data 2420 by
It is fed in map generating unit 2410.Then, map generating unit 2410 is run to generate map output 2426.
Unsupervised method can also be implemented to generate map.Such process may rely on yardstick correlated characteristic to retouch
State the contextual information of each point map.They may also rely on deep learning come design feature conversion, so as to point map
Feature is used together.The totality of the Feature Conversion generated by deep learning is used to encode point map contextual information.
Then the assets membership qualification put can be based on the feature changed by deep learning algorithm.Another method is to surround to be based on class
The study of journey, wherein with course description assets, then calculating figure learning assets.When asset of interest is in shape and property
On be rule and when not showing very big space complexity, this method can be effective.
Using these Learning Schemes, neutral net is often trained in first step, is then applied to geographical space
The remainder of data is for extraction map.
Therefore, machine learning techniques can help to optimize and refine to calculate figure.It can be planned manually using the above method
Or learn these figures.Parameter search part, which is used for accuracy, improves and reduces wrong report and negative.In this step, can modulate
Calculate figure all various parameters (from region of interest to the parameter of each function, the number of the feature used into grader
Amount and essence), and monitor output.By using searching method, the optimum performance that can find parameter is combined and is applied to
Remaining data.This step assumes the availability of the semantic map of noted earlier.
When calculating figure is refined acceptable performance level, they are used directly in vehicle.This will be corresponding
Vehicle is streamed to from cloud in wisdom, rather than more conventional data stream to cloud system from home environment.Use geographical space number
According to the absolute size of sensing data may make us hanging back.Therefore, in some embodiments, the biography locally obtained
Sensor data (data such as obtained by vehicle installing type sensor) are collected via local computing resource, wherein only gathering
Information and/or the subset of content of extraction be sent back to remote data system.For example, with data storage/pretreatment component
1220th, the suitable resource of processing unit 1240 and data analysis institution 1250 can be realized in vehicle, with from onboard sensor
System extracts semantic map datum.The calculating figure described with those above for the implementation in the processing structure based on cloud
Similar calculating figure can be optimized and tested in machine learning framework, while provide chance for local vehicle-mounted realization.
Vehicle can be used as Distributed Computing Platform by such embodiment, constantly update the content of the map of central maintenance, simultaneously
The data of most of remote sense are instead consumed, rather than its whole is streamed to the system based on cloud in center.
Although machine learning implementation described herein can significantly speed up the exploitation of new figure, to map new feature
And assets, but learn the problem of exercise there may come a time when by the shortage of training data and on accuracy.These problems
Consequence may include overfitting and UPS upper performance score.When the amount of training data is limited, study routine may make the property of figure
Available little data can be heavily biased towards, makes it easy to the new feelings being trained in introducing study routine for it
Fail during condition.On performance, the map for creating training data is typically error-prone manual processes.As such, work as training data
When itself is not exclusively accurate, the figure of generation is also inaccurate.For example, if GIS analysts are in its map manually generated
Obtain only 80% assets accuracy, then will all be very difficult to more than 80% based on any figure that the data are trained
Accuracy threshold value.
In order to solve these problems, simulated environment can be utilized.In simulated environment, programming life is arranged with substantial amounts of parameter
Into map, to replicate the region on earth surface and the variability of terrestrial reference.Then generate threedimensional model from map and carry out ray
Tracking, to create point cloud with True Data collection similar mode as far as possible.Because the position of each assets is that priori is known
, so being then able to extract perfect map from a cloud.The variability of data and exist perfectly for each cloud
The fact that face fact, substantially increases the scope and its accuracy for calculating figure.It also offers one kind to understand current meter
Calculate the restricted mechanism of example.
However, no matter figure has been trained to how many, and it experienced how many test case, and automatic map extraction is never
Can be preferable.For this reason, manual quality control (QC) step can be introduced to help to find out any problem.In order to avoid
QC must be performed to whole map, level of confidence can be generated during cartography.The water-glass diagram shape is from ground
How is the confidence level of characteristic aspect desired by extraction in figure.Then QC can be performed to the region of lowest confidence percentile.
Quality control can perform in many ways.Similar to semantic map is created, GIS analysts can use traditional
Visualization tool, and with the original survey data of ground map combining automatically extracted.Then it can identify and correct any difference.For
QC another method is to carry out mass-rent to workload between multiple agents online.Because each in these agents
May be not fully skilled to semantic map building, so QC work will need to be replicated.It may then pass through each QC results
Confirm or deny it is assumed that and drawing final conclusion by enough experiments.
Obtain QC results come strengthen calculate figure be important.When a discrepancy is detected, can be problematic using including
The world of the new simulation of test case.The further retraining of figure can contemplate the service condition in the work in future.
Although for the purpose being aware and understood, some embodiments have been describe in detail herein, above
Description and the accompanying drawing only explanation and illustration present invention, and the invention is not restricted to this.It will be recognized that those skilled in the art exist
They know in the case of present disclosure by can in the case where not departing from the scope of any claim it is public to institute herein
The content opened makes these and other modifications and variations.
Claims (25)
1. a kind of device of assets in identification point cloud survey data, described device include:
For the front end component of receiving point cloud data set, the front end component can access via digital communications network;
Data storage part, the data storage part stores the cloud data collection, and the cloud data collection is segmented
For multiple data blocks;
Processing unit including computing cluster, the processing unit receive the data block of streaming from the data storage part, and
And one or more analysis institutions are applied to each data block with withdrawal of assets information;And
The assets information extracted from the data analysis mechanism is combined into output ground by map generator, the map generator
Figure.
2. device according to claim 1, wherein, each data block includes one or more points cloud data segment.
3. device according to claim 2, wherein, each segment is included in the gravitational vectors Longitudinal extending along the earth
Rectangle row in cloud data subset.
4. device according to claim 3, wherein, each data block includes and is optimized to realize target data block size
Multiple continuous segments.
5. device according to claim 1, wherein, the map generator also includes annotation integrity verification device, described
Annotate integrity verification device the assets information in output map and the one or more corresponding to public home environment is previously defeated
The assets information gone out in map is compared, and is notified with being generated when detecting difference.
6. device according to claim 1, in addition to:
Compression mechanism, the compression mechanism are run to be pressed before the cloud data is stored into the data storage part
Contract the cloud data;And
Mechanism is decompressed, the decompression mechanism is run to decompress institute before the processing unit applies the analysis institution
State cloud data;
Thus, the compression mechanism modulates its compression ratio so that from the data retrieval rate of the data storage part and by described
The PDR balance that processing unit can be realized.
7. a kind of vehicle locating device, including:
The gps receiver of vehicle is attached to, the gps receiver provides the first geographical position of the vehicle;
The local map buffer resided in the vehicle, the local map of the local map buffer storage assets are right
In each assets, the local map includes position, the property associated with the assets and one relative to other assets
Or multiple relations;
One or more home environment sensors, one or more of home environment sensors be arranged on the vehicle on so that
The data associated with the home environment of the du vehicule can be gathered by obtaining;
One or more vehicle computers, the vehicle computer receive first geographical position from the gps receiver, with
The record associated with the assets of the previous drafting in first geographic vicinity is retrieved from the local map buffer;
The characteristic extracting component realized by the vehicle computer, the characteristic extracting component receive the home environment sensor
Data, with identify and position at present the du vehicule observe assets;And
The position realized by the vehicle computer refines part, and the position refinement part will come from the characteristic extracting component
Observe assets mark and position with the assets information retrieved from the local map buffer compared with, with determination
The current state of the vehicle.
8. vehicle locating device according to claim 7, wherein, the current state of the vehicle includes vehicle position
Put.
9. vehicle locating device according to claim 8, wherein, the current state of the vehicle also includes vehicle speed
Degree and direct of travel.
10. vehicle locating device according to claim 7, in addition to wireless vehicle communications equipment, via the wireless vehicle
Communication equipment, the local map buffer can download local map data during vehicle is run from remote data base.
11. vehicle locating device according to claim 7, wherein, one or more of home environment sensors include
One or more of below:LIDAR sensor, digital camera and radar sensor.
12. vehicle locating device according to claim 7, wherein, home environment attribute includes direction board, wayside security ties
One or more of structure and semaphore signal.
13. vehicle locating device according to claim 7, wherein, it is associated with the home environment of the du vehicule
The data of the home environment sensor include three dimensional point cloud.
14. vehicle locating device according to claim 7, wherein, the vehicle is train.
15. vehicle locating device according to claim 14, wherein, the current state of the vehicle includes track mark
Know.
16. vehicle locating device according to claim 7, in addition to interface unit, by the interface unit, the car
The current state can be sent to one or more vehicle control systems.
17. vehicle locating device according to claim 10, in addition to:
Map audits part, and the map examination & verification part is identified between local map and the observed assets of the assets
Difference, and the difference is output to the vehicle communication device for being transferred to the remote data base.
18. vehicle locating device according to claim 17, wherein, the map examination & verification part includes missing assets detection
Device, the missing assets detector identification are present in the assets observed and are not present in the local map of the assets
Interior assets, or identify the money for being not present in being present in the assets observed in the local map of the assets
Production.
19. vehicle locating device according to claim 17, wherein, the map examination & verification part includes assets alteration detection
Device, the interior assets with instruction harm or the characteristic distorted of the assets alteration detection device identification assets observed, institute
State characteristic from and the local map of the assets in assets it is associated characteristic it is different.
20. vehicle locating device according to claim 10, wherein, the vehicle is suitable to advance on the railroad track;Institute
Stating device also includes:
Track headroom evaluation means, the track headroom evaluation means receive the first assets of instruction from the characteristic extracting component
The information of position, the track headroom evaluation means are by first asset identification for barrier and via the vehicle communication
Equipment reports the position of the barrier to back-end server.
21. a kind of be used to audit map datum by one or more gateway servers to safeguard the side of the map in database
Method, it the described method comprises the following steps:
The request to map datum from vehicle is received, the vehicle has home environment sensor and local map caching
Device;
Map Data Transmission to the vehicle, the map datum are included into assets information, the assets in response to the request
Information includes mark, feature and the position of one or more assets;And
Indicated from the vehicle receiver between the map datum and the information detected by the home environment sensor of the vehicle
One or more difference report;And
Based on database described in the information updating in the report.
22. according to the method for claim 21, wherein, the report includes being examined by the home environment sensor of the vehicle
The mark of assets that are surveying and being not present in the database and position.
23. according to the method for claim 21, wherein, the report includes being present in the database but not by institute
State the mark for the assets that the home environment sensor of vehicle detects.
24. according to the method for claim 21, wherein, the assets information includes one or more asset characters;And
Wherein described report includes asset character and the money of the home environment sensor detection by the vehicle with the database
Produce the difference between characteristic.
25. according to the method for claim 21, wherein, the vehicle is train, and described refer to from the vehicle receiver
Show the map datum and the report by one or more difference between the information of the home environment sensor detection of the vehicle
The step of announcement, includes:
Receive report of the instruction relative to the barrier headroom in the path of the train.
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US201562105696P | 2015-01-20 | 2015-01-20 | |
US62/105,696 | 2015-01-20 | ||
PCT/US2016/014196 WO2016118672A2 (en) | 2015-01-20 | 2016-01-20 | Real time machine vision and point-cloud analysis for remote sensing and vehicle control |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107533630A true CN107533630A (en) | 2018-01-02 |
Family
ID=56417920
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201680006431.6A Pending CN107533630A (en) | 2015-01-20 | 2016-01-20 | For the real time machine vision of remote sense and wagon control and put cloud analysis |
Country Status (4)
Country | Link |
---|---|
EP (1) | EP3248140A4 (en) |
JP (1) | JP2018508418A (en) |
CN (1) | CN107533630A (en) |
WO (1) | WO2016118672A2 (en) |
Cited By (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108416808A (en) * | 2018-02-24 | 2018-08-17 | 斑马网络技术有限公司 | The method and device of vehicle reorientation |
CN108648150A (en) * | 2018-05-10 | 2018-10-12 | 句容康泰膨润土有限公司 | A kind of image split-joint method |
CN109948661A (en) * | 2019-02-27 | 2019-06-28 | 江苏大学 | A kind of 3D vehicle checking method based on Multi-sensor Fusion |
CN110481601A (en) * | 2019-09-04 | 2019-11-22 | 深圳市镭神智能系统有限公司 | A kind of rail detection system |
CN110968588A (en) * | 2018-10-01 | 2020-04-07 | 通用汽车环球科技运作有限责任公司 | Decentralized distributed map using blockchains |
CN111098862A (en) * | 2018-10-29 | 2020-05-05 | 通用汽车环球科技运作有限责任公司 | System and method for predicting sensor information |
CN111104849A (en) * | 2018-10-29 | 2020-05-05 | 安波福技术有限公司 | Automatic annotation of environmental features in a map during navigation of a vehicle |
CN111145166A (en) * | 2019-12-31 | 2020-05-12 | 北京深测科技有限公司 | Safety monitoring method and system |
CN111198904A (en) * | 2018-11-16 | 2020-05-26 | 千寻位置网络有限公司 | Data processing method and device and processing system |
CN111223107A (en) * | 2019-12-31 | 2020-06-02 | 武汉中海庭数据技术有限公司 | Point cloud data set manufacturing system and method based on point cloud deep learning |
CN111427985A (en) * | 2020-03-25 | 2020-07-17 | 北京小马智行科技有限公司 | Map updating method and device, storage medium and processor |
CN111806520A (en) * | 2019-04-12 | 2020-10-23 | 泰雷兹管理与服务德国股份有限公司 | Method for the secure, autonomous determination of position information of a train on a track |
CN111935642A (en) * | 2019-05-13 | 2020-11-13 | 北京地平线机器人技术研发有限公司 | Positioning method and device of movable equipment |
CN111976789A (en) * | 2020-05-11 | 2020-11-24 | 香港铁路有限公司 | Train-mounted system and method for determining safe speed and position of train |
CN112019857A (en) * | 2019-05-29 | 2020-12-01 | 诺基亚技术有限公司 | Method and apparatus for storage and signaling of compressed point clouds |
CN112132896A (en) * | 2020-09-16 | 2020-12-25 | 北京埃福瑞科技有限公司 | Trackside equipment state detection method and system |
CN112166304A (en) * | 2018-04-03 | 2021-01-01 | 祖克斯有限公司 | Error detection of sensor data |
CN112241129A (en) * | 2019-07-17 | 2021-01-19 | 中车长春轨道客车股份有限公司 | Control system, method, apparatus, computer device, and storage medium |
CN112313095A (en) * | 2018-04-27 | 2021-02-02 | 大陆汽车系统公司 | Apparatus and method for determining the center of a trailer hitch coupler |
CN112441075A (en) * | 2019-08-30 | 2021-03-05 | 比亚迪股份有限公司 | Rail transit external environment sensing system and method and rail transit equipment |
CN112839857A (en) * | 2018-09-14 | 2021-05-25 | 西门子交通有限公司 | Automated vehicle-side control system for a rail vehicle |
CN113514030A (en) * | 2021-07-12 | 2021-10-19 | 石家庄铁道大学 | Ballast bed section detection method and system |
CN113924459A (en) * | 2018-12-14 | 2022-01-11 | 北京航迹科技有限公司 | Dynamic sensor range detection for vehicle navigation |
CN114829227A (en) * | 2019-10-16 | 2022-07-29 | 北伯林顿铁路公司 | Asset auditing system and method |
CN117311283A (en) * | 2023-10-24 | 2023-12-29 | 风凯换热器制造(常州)有限公司 | Workshop running control intelligent monitoring method and system for preassembly body in heat exchanger |
US11999396B2 (en) | 2018-09-14 | 2024-06-04 | Siemens Mobility GmbH | Automated on-vehicle control system for a rail vehicle |
Families Citing this family (41)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR3062103B1 (en) * | 2017-01-20 | 2019-04-05 | Pierre Messulam | SYSTEM FOR AIDING THE DRIVING OF RAILWAY VEHICLES |
US11608097B2 (en) | 2017-02-28 | 2023-03-21 | Thales Canada Inc | Guideway mounted vehicle localization system |
US10430968B2 (en) * | 2017-03-14 | 2019-10-01 | Ford Global Technologies, Llc | Vehicle localization using cameras |
AR112184A1 (en) * | 2017-04-14 | 2019-10-02 | Bayer Cropscience Lp | VEGETATION DETECTION AND ALERT SYSTEM FOR A RAILWAY VEHICLE AND METHOD |
US10599138B2 (en) | 2017-09-08 | 2020-03-24 | Aurora Flight Sciences Corporation | Autonomous package delivery system |
US20190095725A1 (en) * | 2017-09-22 | 2019-03-28 | Aurora Flight Sciences Corporation | Detection System for a Vehicle |
US10426393B2 (en) | 2017-09-22 | 2019-10-01 | Aurora Flight Sciences Corporation | Systems and methods for monitoring pilot health |
US10699135B2 (en) | 2017-11-20 | 2020-06-30 | Here Global B.V. | Automatic localization geometry generator for stripe-shaped objects |
FR3080823B1 (en) * | 2018-05-03 | 2022-04-29 | Thales Sa | INTEGRATED AND AUTONOMOUS LOCATION SYSTEM OF A TRAIN IN A RAILWAY NETWORK REPOSITORY |
EP3802267A1 (en) * | 2018-05-25 | 2021-04-14 | Bayer CropScience LP | System and method for vegetation management risk assessment and resolution |
US11136120B2 (en) | 2018-10-05 | 2021-10-05 | Aurora Flight Sciences Corporation | Ground operations for autonomous object pickup |
CN109286946B (en) * | 2018-11-13 | 2021-03-09 | 山东建筑大学 | Mobile communication indoor wireless network optimization method and system based on unsupported positioning |
US11346950B2 (en) * | 2018-11-19 | 2022-05-31 | Huawei Technologies Co., Ltd. | System, device and method of generating a high resolution and high accuracy point cloud |
KR102043405B1 (en) * | 2019-06-20 | 2019-11-12 | 주식회사 첨단공간정보 | System of image processing and editing based on GIS |
JP7343153B2 (en) * | 2019-08-08 | 2023-09-12 | 独立行政法人自動車技術総合機構 | train position detection device |
DE102019212010A1 (en) * | 2019-08-09 | 2021-02-11 | Siemens Mobility GmbH | Track detection with lidar |
WO2021067935A1 (en) * | 2019-10-04 | 2021-04-08 | Waymo Llc | Spatio-temporal embeddings |
US11281917B2 (en) * | 2019-10-31 | 2022-03-22 | Aptiv Technologies Limited | Multi-domain neighborhood embedding and weighting of point cloud data |
CN111046765B (en) * | 2019-12-02 | 2023-07-14 | 北京深测科技有限公司 | Dangerous early warning method and system for high-speed rail |
US20210223363A1 (en) * | 2020-01-16 | 2021-07-22 | Outsight SA | Object detection on a path of travel and obstacle detection on railway tracks using free space information |
CN113677583B (en) * | 2020-01-21 | 2023-08-15 | 深圳元戎启行科技有限公司 | Graph calculation-based vehicle driving data processing method and device and computer equipment |
CN111091700A (en) * | 2020-03-23 | 2020-05-01 | 北京全路通信信号研究设计院集团有限公司 | Comprehensive configuration intelligent terminal for track circuit outdoor monitoring equipment |
JP7170937B2 (en) * | 2020-03-30 | 2022-11-14 | 三菱電機株式会社 | Data extraction device, data extraction method and data extraction program |
KR20230014724A (en) * | 2020-05-22 | 2023-01-30 | 프로파운드 포지셔닝 인코포레이티드 | Vehicle localization system and method |
EP3919344A1 (en) * | 2020-06-05 | 2021-12-08 | Mitsubishi Electric R&D Centre Europe B.V. | Method for geolocating an interference source in a communication-based transport system |
US11048717B1 (en) | 2020-06-16 | 2021-06-29 | Geotab Inc. | Dataset simplification of N-dimensional signals captured for asset tracking |
CN111832548B (en) * | 2020-06-29 | 2022-11-15 | 西南交通大学 | Train positioning method |
US11609888B2 (en) | 2020-07-31 | 2023-03-21 | Geotab Inc. | Methods and systems for fixed interpolation error data simplification processes for telematics |
US11593329B2 (en) | 2020-07-31 | 2023-02-28 | Geotab Inc. | Methods and devices for fixed extrapolation error data simplification processes for telematics |
US11556509B1 (en) | 2020-07-31 | 2023-01-17 | Geotab Inc. | Methods and devices for fixed interpolation error data simplification processes for telematic |
US20220091266A1 (en) * | 2020-09-18 | 2022-03-24 | Denso International America, Inc. | Systems and methods for enhancing outputs of a lidar |
CN112184869A (en) * | 2020-10-09 | 2021-01-05 | 北京理工大学 | Point cloud simplification method for keeping geometric features based on absolute Gaussian curvature estimation |
CN112363837A (en) * | 2020-11-19 | 2021-02-12 | 北京航天泰坦科技股份有限公司 | Cluster environment-based point cloud data parallel processing method, device, equipment and storage medium |
US11546395B2 (en) | 2020-11-24 | 2023-01-03 | Geotab Inc. | Extrema-retentive data buffering and simplification |
US11838364B2 (en) * | 2020-11-24 | 2023-12-05 | Geotab Inc. | Extrema-retentive data buffering and simplification |
IT202000030677A1 (en) * | 2020-12-15 | 2022-06-15 | Ecosud S R L | DYNAMIC DEVICE AND RELATED METHOD FOR CARTOGRAPHIC INSPECTION AND FOR THE TRANSFER OF KNOW-HOW |
DE102020216014A1 (en) * | 2020-12-16 | 2022-06-23 | Siemens Mobility GmbH | Method of controlling a device |
CN112669461B (en) * | 2021-01-07 | 2024-01-26 | 中煤航测遥感集团有限公司 | Airport clearance safety detection method and device, electronic equipment and storage medium |
CN113536883B (en) * | 2021-03-23 | 2023-05-02 | 长沙智能驾驶研究院有限公司 | Obstacle detection method, vehicle, apparatus, and computer storage medium |
CN113870123B (en) * | 2021-08-19 | 2023-01-03 | 中国铁路设计集团有限公司 | Automatic detection method for contact net leading height and pulling value based on vehicle-mounted mobile laser point cloud |
FR3140506A1 (en) * | 2022-09-29 | 2024-04-05 | Thales | Method for managing network reconfigurations of a hybrid on-board/ground communication system of a railway rolling vehicle during a journey |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101697229A (en) * | 2009-10-30 | 2010-04-21 | 宁波大学 | Method for extracting region of interest of medical image |
CN102799898A (en) * | 2012-06-28 | 2012-11-28 | 浙江大学 | Efficient target identification processing method of high-resolution remote sensing image in context |
US20130096886A1 (en) * | 2010-03-31 | 2013-04-18 | Borys Vorobyov | System and Method for Extracting Features from Data Having Spatial Coordinates |
WO2014098951A1 (en) * | 2012-12-21 | 2014-06-26 | Wabtec Holding Corp. | Track data determination system and method |
JP2014185947A (en) * | 2013-03-25 | 2014-10-02 | Geo Technical Laboratory Co Ltd | Image photographing method for three-dimensional restoration |
US9245170B1 (en) * | 2010-02-24 | 2016-01-26 | The Boeing Company | Point cloud data clustering and classification using implicit geometry representation |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4086994B2 (en) * | 1999-02-08 | 2008-05-14 | 株式会社デンソー | Image data supply device and image compression device |
JP2006321277A (en) * | 2005-05-17 | 2006-11-30 | Hitachi Ltd | Railroad facilities information providing system, its method, and its program |
US8605947B2 (en) * | 2008-04-24 | 2013-12-10 | GM Global Technology Operations LLC | Method for detecting a clear path of travel for a vehicle enhanced by object detection |
JP2010202017A (en) * | 2009-03-03 | 2010-09-16 | Mitsubishi Electric Corp | Data analysis device and method, and program |
US20140379254A1 (en) * | 2009-08-25 | 2014-12-25 | Tomtom Global Content B.V. | Positioning system and method for use in a vehicle navigation system |
US20110216063A1 (en) * | 2010-03-08 | 2011-09-08 | Celartem, Inc. | Lidar triangular network compression |
US9102341B2 (en) * | 2012-06-15 | 2015-08-11 | Transportation Technology Center, Inc. | Method for detecting the extent of clear, intact track near a railway vehicle |
US9221461B2 (en) * | 2012-09-05 | 2015-12-29 | Google Inc. | Construction zone detection using a plurality of information sources |
DE102013104088A1 (en) * | 2013-04-23 | 2014-10-23 | Deutsches Zentrum für Luft- und Raumfahrt e.V. | Method for automatically detecting characteristic elements, in particular a level crossing, and device therefor |
-
2016
- 2016-01-20 WO PCT/US2016/014196 patent/WO2016118672A2/en active Application Filing
- 2016-01-20 CN CN201680006431.6A patent/CN107533630A/en active Pending
- 2016-01-20 EP EP16740714.7A patent/EP3248140A4/en not_active Withdrawn
- 2016-01-20 JP JP2017556798A patent/JP2018508418A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101697229A (en) * | 2009-10-30 | 2010-04-21 | 宁波大学 | Method for extracting region of interest of medical image |
US9245170B1 (en) * | 2010-02-24 | 2016-01-26 | The Boeing Company | Point cloud data clustering and classification using implicit geometry representation |
US20130096886A1 (en) * | 2010-03-31 | 2013-04-18 | Borys Vorobyov | System and Method for Extracting Features from Data Having Spatial Coordinates |
CN102799898A (en) * | 2012-06-28 | 2012-11-28 | 浙江大学 | Efficient target identification processing method of high-resolution remote sensing image in context |
WO2014098951A1 (en) * | 2012-12-21 | 2014-06-26 | Wabtec Holding Corp. | Track data determination system and method |
JP2014185947A (en) * | 2013-03-25 | 2014-10-02 | Geo Technical Laboratory Co Ltd | Image photographing method for three-dimensional restoration |
Cited By (39)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108416808A (en) * | 2018-02-24 | 2018-08-17 | 斑马网络技术有限公司 | The method and device of vehicle reorientation |
CN108416808B (en) * | 2018-02-24 | 2022-03-08 | 斑马网络技术有限公司 | Vehicle repositioning method and device |
CN112166304A (en) * | 2018-04-03 | 2021-01-01 | 祖克斯有限公司 | Error detection of sensor data |
CN112313095A (en) * | 2018-04-27 | 2021-02-02 | 大陆汽车系统公司 | Apparatus and method for determining the center of a trailer hitch coupler |
CN108648150A (en) * | 2018-05-10 | 2018-10-12 | 句容康泰膨润土有限公司 | A kind of image split-joint method |
CN112839857A (en) * | 2018-09-14 | 2021-05-25 | 西门子交通有限公司 | Automated vehicle-side control system for a rail vehicle |
US11999396B2 (en) | 2018-09-14 | 2024-06-04 | Siemens Mobility GmbH | Automated on-vehicle control system for a rail vehicle |
CN110968588A (en) * | 2018-10-01 | 2020-04-07 | 通用汽车环球科技运作有限责任公司 | Decentralized distributed map using blockchains |
CN111104849A (en) * | 2018-10-29 | 2020-05-05 | 安波福技术有限公司 | Automatic annotation of environmental features in a map during navigation of a vehicle |
US11774261B2 (en) | 2018-10-29 | 2023-10-03 | Motional Ad Llc | Automatic annotation of environmental features in a map during navigation of a vehicle |
CN111098862A (en) * | 2018-10-29 | 2020-05-05 | 通用汽车环球科技运作有限责任公司 | System and method for predicting sensor information |
CN111198904A (en) * | 2018-11-16 | 2020-05-26 | 千寻位置网络有限公司 | Data processing method and device and processing system |
CN113924459B (en) * | 2018-12-14 | 2024-04-16 | 广州沃芽来得灵科技有限公司 | Dynamic sensor range detection for vehicle navigation |
CN113924459A (en) * | 2018-12-14 | 2022-01-11 | 北京航迹科技有限公司 | Dynamic sensor range detection for vehicle navigation |
CN109948661A (en) * | 2019-02-27 | 2019-06-28 | 江苏大学 | A kind of 3D vehicle checking method based on Multi-sensor Fusion |
CN109948661B (en) * | 2019-02-27 | 2023-04-07 | 江苏大学 | 3D vehicle detection method based on multi-sensor fusion |
US11623673B2 (en) | 2019-04-12 | 2023-04-11 | Thales Management & Services Deutschland Gmbh | Method for safely and autonomously determining the position information of a train on a track |
CN111806520A (en) * | 2019-04-12 | 2020-10-23 | 泰雷兹管理与服务德国股份有限公司 | Method for the secure, autonomous determination of position information of a train on a track |
CN111935642A (en) * | 2019-05-13 | 2020-11-13 | 北京地平线机器人技术研发有限公司 | Positioning method and device of movable equipment |
CN112019857A (en) * | 2019-05-29 | 2020-12-01 | 诺基亚技术有限公司 | Method and apparatus for storage and signaling of compressed point clouds |
CN112241129A (en) * | 2019-07-17 | 2021-01-19 | 中车长春轨道客车股份有限公司 | Control system, method, apparatus, computer device, and storage medium |
CN112441075A (en) * | 2019-08-30 | 2021-03-05 | 比亚迪股份有限公司 | Rail transit external environment sensing system and method and rail transit equipment |
CN110481601A (en) * | 2019-09-04 | 2019-11-22 | 深圳市镭神智能系统有限公司 | A kind of rail detection system |
CN114829227A (en) * | 2019-10-16 | 2022-07-29 | 北伯林顿铁路公司 | Asset auditing system and method |
CN114829227B (en) * | 2019-10-16 | 2024-04-12 | 北伯林顿铁路公司 | Asset auditing system and method |
CN111223107A (en) * | 2019-12-31 | 2020-06-02 | 武汉中海庭数据技术有限公司 | Point cloud data set manufacturing system and method based on point cloud deep learning |
CN111145166B (en) * | 2019-12-31 | 2023-09-01 | 北京深测科技有限公司 | Security monitoring method and system |
CN111145166A (en) * | 2019-12-31 | 2020-05-12 | 北京深测科技有限公司 | Safety monitoring method and system |
CN111427985A (en) * | 2020-03-25 | 2020-07-17 | 北京小马智行科技有限公司 | Map updating method and device, storage medium and processor |
CN111427985B (en) * | 2020-03-25 | 2023-04-11 | 北京小马慧行科技有限公司 | Map updating method and device, storage medium and processor |
CN111976789A (en) * | 2020-05-11 | 2020-11-24 | 香港铁路有限公司 | Train-mounted system and method for determining safe speed and position of train |
WO2021227305A1 (en) * | 2020-05-11 | 2021-11-18 | Mtr Corporation Limited | On-board systems for trains and methods of determining safe speeds and locations of trains |
CN111976789B (en) * | 2020-05-11 | 2023-02-17 | 智慧铁路科技有限公司 | Train-mounted system and method for determining safe speed and position of train |
CN112132896A (en) * | 2020-09-16 | 2020-12-25 | 北京埃福瑞科技有限公司 | Trackside equipment state detection method and system |
CN112132896B (en) * | 2020-09-16 | 2024-05-10 | 北京埃福瑞科技有限公司 | Method and system for detecting states of trackside equipment |
CN113514030A (en) * | 2021-07-12 | 2021-10-19 | 石家庄铁道大学 | Ballast bed section detection method and system |
CN113514030B (en) * | 2021-07-12 | 2022-08-23 | 石家庄铁道大学 | Ballast bed section detection method and system |
CN117311283A (en) * | 2023-10-24 | 2023-12-29 | 风凯换热器制造(常州)有限公司 | Workshop running control intelligent monitoring method and system for preassembly body in heat exchanger |
CN117311283B (en) * | 2023-10-24 | 2024-03-19 | 风凯换热器制造(常州)有限公司 | Workshop running control intelligent monitoring method and system for preassembly body in heat exchanger |
Also Published As
Publication number | Publication date |
---|---|
JP2018508418A (en) | 2018-03-29 |
WO2016118672A3 (en) | 2016-10-20 |
WO2016118672A2 (en) | 2016-07-28 |
EP3248140A4 (en) | 2018-12-05 |
EP3248140A2 (en) | 2017-11-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107533630A (en) | For the real time machine vision of remote sense and wagon control and put cloud analysis | |
US10549768B2 (en) | Real time machine vision and point-cloud analysis for remote sensing and vehicle control | |
US11593344B2 (en) | Updating high definition maps based on age of maps | |
US11988518B2 (en) | Updating high definition maps based on lane closure and lane opening | |
CN110832474B (en) | Method for updating high-definition map | |
US10832502B2 (en) | Calibration for autonomous vehicle operation | |
US11106218B2 (en) | Adaptive mapping to navigate autonomous vehicles responsive to physical environment changes | |
CN104851295B (en) | Obtain the method and system of traffic information | |
US10037689B2 (en) | Apparatus and system to manage monitored vehicular flow rate | |
US20210042500A1 (en) | Automatic detection and positioning of pole-like objects in 3d | |
CN109643367A (en) | Crowdsourcing and the sparse map of distribution and lane measurement for autonomous vehicle navigation | |
CN109415043A (en) | A kind of automatic Pilot traveling planing method based on comfort level | |
US11514276B2 (en) | System and method for detecting and classifying recurrent stops of a vehicle fleet | |
CN107782564A (en) | A kind of automatic driving vehicle evaluation system and method | |
CN109724610A (en) | A kind of method and device of full information real scene navigation | |
CN108974044A (en) | Railroad track assets survey system | |
CN107851125A (en) | The processing of two step object datas is carried out by vehicle and server database to generate, update and transmit the system and method in accurate road characteristic data storehouse | |
JP2019502214A (en) | Adaptive mapping for navigating autonomous vehicles in response to changes in the physical environment | |
CN107850672A (en) | System and method for accurate vehicle positioning | |
CN110990661B (en) | Region of interest access extraction method based on road network constraint and density clustering | |
CN109615851A (en) | A kind of sensing node choosing method in intelligent perception system based on key road segment | |
CN106469339A (en) | A kind of around geographic information application and provide the system of infrastructure service | |
CN102693642A (en) | Variable message sign application system, electronic navigation method and device thereof | |
Liu et al. | Determination of routing velocity with GPS floating car data and webGIS-based instantaneous traffic information dissemination | |
CN116805444A (en) | Distributed named navigation pedestrian driving early warning method, system and device |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180102 |
|
WD01 | Invention patent application deemed withdrawn after publication |