CN108010360A - A kind of automatic Pilot context aware systems based on bus or train route collaboration - Google Patents

Abstract

The invention discloses a kind of automatic Pilot context aware systems based on bus or train route collaboration, it is included in trackside sensor, trackside processing unit and the trackside communication unit set on the section by setpoint distance division.Wherein, trackside sensor includes camera, laser radar and microwave radar, for gathering the road information of road；Trackside processing unit is used for the road information for receiving the collection of trackside sensor, it is transformed into after processing is identified in unified world coordinate system, then transformed data fusion is labeled in high-precision map, semantization road environment information is formed after splicing to the high-precision maps of adjacent segments, is transmitted to trackside communication unit；Trackside communication unit, for being sent semantization road environment information by wireless communication technique.The present invention is by trackside sensor, and on the basis of original high-precision map, according to the road behavioral characteristics data collected, real-time update high accuracy map, global context perception is provided for automatic driving vehicle safety traffic.

Description

A kind of automatic Pilot context aware systems based on bus or train route collaboration

Technical field

The invention belongs to intelligent network to join technical field of transportation, more particularly to a kind of automatic Pilot environment based on bus or train route collaborationSensory perceptual system.

Background technology

Forefront technology of the automatic Pilot technology as current intelligent transportation field, includes environment sensing, the road of vehiclePlan and decision-making and vehicle motion control in footpath.At this stage, with the development of artificial intelligence, deep learning process chip, threeAutomatic Pilot technology is obtained for rapid development.However, to realize complete perception of the vehicle to surrounding enviroment, each is driven automaticallySailing vehicle all needs to be equipped with complicated, expensive, a variety of sensor devices, causes integral vehicle cost to remain high, according to autonomous type routeEvolution, autonomous driving vehicle commercialization process are slow.

The environment perception technology of autonomous driving vehicle mainstream is merged by a variety of onboard sensors, and combined high precisionFigure realizes the complete perception to peripheral path environment, this kind of method has some following shortcoming：

(1) vehicle sensing range is had a great influence by sensor performance, current to perceive 200 meters or so of ultimate range, for heightThe vehicle sensing range smaller of speed traveling；

(2) sensing range is vulnerable to barrier influence, it is difficult to realizes the complete perception to surrounding enviroment, current vehicle is autonomousThe cognition technology of formula can not realize the perception front situation that clears the jumps, it is impossible to solve the driving safety in the case of field obscurationProblem；

(3) context aware systems of each automatic driving vehicle independence bring challenges automatic Pilot formation traveling, automaticallyThe sensing data that platooning is difficult to merge all vehicles in fleet when driving is driven, causes fleet systems poor robustness, poleEasily disintegrate；

(4) the problems such as current high-precision map is relatively low there are renewal frequency, and collection of material cost is higher, it is difficult to meet automaticThe actual demand of driving.

The content of the invention

It is existing for making up the object of the present invention is to provide a kind of automatic Pilot context aware systems based on bus or train route collaborationThe deficiency of the context aware systems of automatic driving vehicle independence under technical conditions.

To achieve these goals, technical solution of the present invention is as follows：

A kind of automatic Pilot context aware systems based on bus or train route collaboration, the automatic Pilot environment based on bus or train route collaborationSensory perceptual system, is included in the trackside sensor set on the section by setpoint distance division, trackside processing unit and trackside communicationUnit, wherein：

The trackside sensor includes camera, laser radar and microwave radar, for gathering the road information of road；

The trackside processing unit, for receiving the road information of trackside sensor collection, is changed after processing is identifiedInto unified world coordinate system, then transformed data fusion is labeled in high-precision map, to the high-precision of adjacent segmentsDegree map forms semantization road environment information after being spliced, and is transmitted to trackside communication unit；

The trackside communication unit, for being sent semantization road environment information by wireless communication technique.

A kind of implementation of the present invention, the camera gather section video image, identify the thing in video imageBody species, shape, positional information.Image recognition work is transferred into camera to complete, the meter of trackside processing unit can be mitigatedCalculation amount.

The trackside processing unit is receiving the road information of trackside sensor collection, and unified generation is transformed into after being identifiedWhen in boundary's coordinate system, following operation is performed：

All trackside sensors are added to unified world coordinate system；

The kind of object of camera acquisition, shape, positional information are transformed into unified world coordinate system；

The laser point cloud being reflected back that laser radar obtains is received, three-dimensionalreconstruction is carried out to surrounding enviroment, is transformed into unificationIn world coordinate system；

The microwave being reflected back that microwave radar obtains is received, identification has clear and corresponding position, is transformed into unificationIn world coordinate system.

Another implementation of the present invention, the trackside processing unit are receiving the road letter of trackside sensor collectionBreath, when being transformed into after being identified in unified world coordinate system, performs following operation：

All trackside sensors are added to unified world coordinate system；

The video image of camera collection is received, kind of object in video image, shape, position is identified, is transformed intoIn unified world coordinate system；

The laser point cloud being reflected back that laser radar obtains is received, three-dimensionalreconstruction is carried out to surrounding enviroment, is transformed into unificationIn world coordinate system；

The microwave being reflected back that microwave radar obtains is received, identification has clear and corresponding position, is transformed into unificationIn world coordinate system.

Further, the trackside sensor, further includes weather detector.The data of weather detector collection, are also transmittedTo trackside processing unit, the second layer or third layer of high-precision map can be placed on after processing.

Further, the trackside communication unit is additionally operable to receive the road status information that administrative center issues, and is sent toTrackside processing unit；

The trackside processing unit, is additionally operable to receive the road status information that trackside communication unit is sent, is labeled in high-precisionSpend in map.

High-precision map described in the present embodiment includes four layers, and first layer is the static high-precision cartographic information of default road,The second layer is road status information layer, and third layer is road dynamic environment Information Level, and the 4th layer is road traffic state Information Level.

Further, it is described that transformed data fusion is labeled in high-precision map, including：

The road status information received by trackside communication unit is labeled in the second layer；

The road information obtained by trackside sensor is labeled in third layer；

According to the road status information received by trackside communication unit and the road information obtained by trackside sensor,Fusion generation macro-traffic flow condition information, is labeled in the 4th layer.

Further, semantization road is formed after the trackside processing unit splices the high-precision map of adjacent segmentsRoad environmental information, performs following operation：

The high-precision map of front and rear adjacent segments is spliced respectively, generates the map of two splicings, then two are spelledThe map connect forms semantization road environment information.

Further, the camera is laid in above road on rod piece with microwave radar, and the laser radar is laid inOn the rod piece of road center dividing strip, the trackside processing unit, trackside communication unit are located on the rod piece of trackside.

A kind of automatic Pilot context aware systems based on bus or train route collaboration proposed by the present invention, by disposing a variety of sensingsDevice, gathers road information, is sent after trackside processing unit processes by trackside communication unit, is received and is used for by car-mounted terminalThe environment sensing of vehicle.Extension or the sensory perceptual system for substituting automatic driving vehicle, peripheral ring is perceived for existing onboard sensorThe deficiency of border scheme, as sensing range is shorter and is easily blocked, vehicle-mounted camera is easily influenced by opposite vehicle high beam, is passed throughThe sensor for being laid in diverse location realizes the convergence of large range of road multidate information, is realized over long distances from " god visual angle "The complete perception of surrounding enviroment.The making of current high accuracy map relies primarily on measurement vehicle and draws, and renewal frequency is difficult to meetThe requirement of automated driving system.This programme can be by trackside sensor, and on the basis of original high-precision map, foundation collectsRoad behavioral characteristics data, real-time update high accuracy map, such as vehicle, pedestrian or other barriers.For automatic driving vehicleSafety traffic provides global context perception.This programme can realize the perception to a wide range of interior road environment, automatic Pilot carTeam can be based on itself formation position Auto-matching, realize the complete perception to all vehicle-surroundings environment in formation, avoid because lackingFew surrounding enviroment information can not dynamically adjust formation, it is impossible to reach the maximized problem of formation efficiency.

Brief description of the drawings

Fig. 1 is that automatic Pilot context aware systems of the present invention based on bus or train route collaboration dispose schematic diagram；

Fig. 2 splices schematic diagram for section of the present invention.

Embodiment

Technical solution of the present invention is described in further details with reference to the accompanying drawings and examples, following embodiments are not formedLimitation of the invention.

The rapid development of the communication technology, which makes between vehicle, vehicle communicates with low time delay between infrastructure becomes reality.TraditionDSRC and the short-range communication technology such as the LTE-V2X that develops rapidly at this stage have been able to realize between vehicle, vehicle and basisThe communication of less than 100 milliseconds time delays between facility, following 5G epoch are up to 1 millisecond of time delay.Short distance, the communication skill of low time delayThe realizing route for developing into automatic Pilot of art provides a kind of new thinking, and the technical program is perceived based on trackside constructing environmentMultiple sensors mounted on a vehicle are transferred to trackside end by system, can effectively solve current automatic driving vehicle costIt is higher, a series of problems, such as environment sensing ability is limited.

A kind of automatic Pilot context aware systems based on bus or train route collaboration of the present embodiment, as shown in Figure 1, being included in by settingTrackside sensor, trackside processing unit and the trackside communication unit set on the section of distance division.

Traffic route is divided into multiple sections, such as 400 meters of each section by the present embodiment according to the distance of setting, specificallyDivision can be variant according to route topography difference, substantially when the length in section is determined according to the performance of each sensor, withEnsure covering of each sensor to the section.

Wherein, trackside sensor mainly includes camera, laser radar, microwave radar, weather detector etc., according to each biographySensor performance, required according to Multi-sensor Fusion algorithm it is uniformly distributed at trackside end, to ensure all standing of the sensor to section.

For example, camera is laid in above road on rod piece with microwave radar, based on camera performance, layout density is horizontal strokeTo every one, 2 track, a row are laid in longitudinal direction for every 200 meters.For ensure camera view covering, rod piece height can properly increase to8 meters.

Laser radar is laid in road center dividing strip, apart from 100 meters of camera rod piece, using low harness multilasered optical radarArray, it is uniformly distributed on 2 meters of rod pieces of height.

Weather detector is laid in trackside, predominantly detects the meteorological datas such as visibility, wind-force.

Each sensor detection range has overlapping to a certain degree so that sensing data splices with merging.Each section trackside is eachSensing data is based on Ethernet-Aggregation to trackside processing unit.The present embodiment can be to avoid using single using multiple sensorsSensor gathers the influence of information inaccuracy, and only merging this multiple sensors information could meet automatic driving vehicle to accurateThe requirement of environment sensing.

The present embodiment trackside processing unit, trackside communication unit are located on the rod piece of trackside, and trackside processing unit is insertionThe three-dimensional extraction algorithm of formula SoC chip, built-in multi-intelligence algorithm, including the image recognition algorithm of view-based access control model, view-based access control model,Three-dimensionalreconstruction algorithm based on laser radar cloud atlas, the Tracking Recognition algorithm based on microwave and the splicing based on multisensor are meltedHop algorithm etc..Its installation position is among section, density is every one, 400 meters of sections.What deserves to be explained is to mitigate centreThe calculation amount of chip is managed, overall calculation framework can use hybrid framework, i.e. the algorithm integration of view-based access control model is taken the photograph to head end is imagedAfter the primary image processing such as video image that head gathers converted by image, noise reduction, the image recognition based on deep learning is calculatedMethod and the three-dimensional extracting method of view-based access control model obtain the information such as kind of object in video flowing, general shape, position in real time.OtherSensor such as laser radar, microwave radar etc. converge to trackside processing unit and are calculated.

It is easily understood that video camera itself can realize the image recognition algorithm of embedded view-based access control model, view-based access control model at presentThree-dimensional extraction algorithm, identify kind of object in video image, shape, positional information.Image recognition work is transferred to take the photographCompleted as head, the calculation amount of trackside processing unit can be mitigated.

The present embodiment trackside processing unit is changed after the data of each trackside sensor are received after processing is identified one by oneInto unified world coordinate system, specifically：

For the video image of camera collection, convert by image, after the primary image processing such as noise reduction, based on depthThe image recognition algorithm of habit and the three-dimensional extracting method of view-based access control model obtain in real time kind of object in video flowing, general shape,The information such as position (or being identified by camera oneself), it is three-dimensional to obtain corresponding object (such as vehicle) based on itself scene libraryInformation, and to its locating and tracking.Then the unification of obtained data is used into UTM coordinate systems to world coordinate system, the present embodiment.

For laser radar, collection is the laser point cloud being reflected back, and trackside processing unit is after the processing such as noise reduction, baseThe accurate three-D profile of barrier and positional information are drawn in the three-dimensionalreconstruction algorithm process point cloud chart of laser radar cloud atlas, it is rightSurrounding enviroment carry out three-dimensionalreconstruction.Then the unification of obtained data is used into UTM coordinate systems to world coordinate system, the present embodiment.

For microwave radar, collection is the microwave being reflected back, Tracking Recognition algorithm of the trackside processing unit based on microwaveIdentification has clear and correspondence position, and the unification of obtained data then is used UTM coordinates to world coordinate system, the present embodimentSystem.

It should be noted that the present embodiment road high accuracy map can use other existing business with pre-production or directlyIndustry map.Each trackside sensor is based on local Coordinate System gathered data, such as camera is based on itself installation site and angle of visibilityDegree, establishes the coordinate system based on camera；Laser radar sensor establishes coordinate system based on self-position；Microwave radar is based on certainlyBody installation site and angle establish coordinate.The data of collection are transmitted to trackside processing unit, trackside processing after gathered dataAll the sensors are added to unified world coordinate system by unit, and UTM coordinate systems can be used to simplify coordinate conversion work.On sittingThe conversion of system is marked, has been the technology of comparative maturity, the present embodiment repeats no more.

By above-mentioned technological means, the sensing data of collection is transformed into unified world coordinate system, generation is highPrecision dynamic road data.Last trackside processing unit is based on each sensing data of multisensor splicing blending algorithm fusion and drawsAccurate road environment data, and semantization road environment information is formed, it is transmitted to trackside communication unit.

What deserves to be explained is to reduce the influence of sensor processing delay, obtained road environment data can be according to detectionThe information such as obtained speed, track do appropriate time compensation, to ensure the real-time accuracy of data.

In sensor array is integrated, the main function of camera is there are object information, auxiliary in identification road environmentPositioning.Laser radar main function is to the accurate three-dimensionalreconstruction of road surrounding enviroment, obtains object dimensional wheel in road environmentWide and positional information.Microwave radar main function is barrier in measurement road environment, and other sensors are supplemented.And gasAs detector acquisition meteorological data, also together issue, be incorporated into by trackside processing unit in high-precision map.

The trackside communication unit of the present embodiment is additionally operable to receive the road status information that administrative center issues, and is sent to tracksideProcessing unit；The trackside processing unit, is additionally operable to receive the road status information that trackside communication unit is sent, is labeled in high-precisionSpend in map.

Administrative center can be traffic control center, or the administrative center of automatic Pilot, after getting road status informationIt is issued to trackside communication unit.

Four layers of the present embodiment high accuracy map point, first layer is the static high-precision cartographic information of default road, by realityThe abstract high accuracy map that road modeling is formed, such as road tomograph, while mark road name, lanes sideTo etc. information；The second layer is road status information, such as speed limit, whether be changeable driveway, whether construct；Third layer is moved for roadThe information such as state environmental information, such as vehicle, pedestrian, the abstract structure of barrier, position, speed, three-dimensional information, the direction of motion,Indeclinable barrier static for a long time is automatically added to base map layer；4th layer is road traffic state information, such as road-section averageSpeed, the degree of crowding, track occupancy situation etc..

Wherein, the present embodiment first layer is the static high-precision cartographic information of default road；Trackside communication unit will be passed throughThe road status information that the administrative center of reception issues is labeled in the second layer；The road information mark that will be obtained by trackside sensorNote is in third layer；Believed according to the road status information received by trackside communication unit and the road obtained by trackside sensorBreath, fusion generation macro-traffic flow condition information, is labeled in the 4th layer.

Trackside processing unit fusion treatment forms data：(1) each object of road abstract, labeling information and markFor note into third layer map, markup information includes object classification, three-dimensional information, position, speed, acceleration-deceleration, direction of motion etc.；(2) according to the road status information received by trackside communication unit and the road information obtained by trackside sensor, fusionMacro-traffic flow condition information is generated, such as congested in traffic situation, track occupancy situation, mark into the 4th layer of map.

And the meteorological data of weather detector monitoring, the second layer can be put into, can also be placed in third layer.

Only it is a kind of implementation it should be noted that the map delamination of above-described embodiment is labeled, art technologyPersonnel can also reduce the level of map or increase the level of map, and the present invention is without limitation.

As shown in Fig. 2, the present embodiment trackside processing unit forms language after splicing to the high-precision map of adjacent segmentsJusticeization road environment information, performs following operation：

The high-precision map of front and rear adjacent segments is spliced respectively, generates the map of two splicings, then two are spelledThe map connect forms semantization road environment information.

Specifically, the opposite map with individually sending a section, preferably, trackside processing unit i is by road for the present embodimentThe map of section i and section i+1 are spliced, and as a map, section i and the map of section i-1 are spliced, as anotherOne map.After carrying out semantization, sent by trackside communication unit., can be with by the way that the map of adjacent segments is splicedComplete road-map is formed, is directly handled easy to car-mounted terminal.

Semantization is finally carried out, the semantization cartographic information for handling completion is to simplify version road environment, and each object is using letterChange module and labeling form shows.

Trackside communication unit can be carried out information exchange with traffic control center, received in management by wired communication modeThe road status information that the heart issues, and be transmitted to trackside processing unit and handled.The also built-in low time delay of trackside communication unit leads toBelieve module, mechanics of communication includes LTE-V2X, DSRC, the mobile low time delay such as edge calculations MEC and future 5G technologies communication skillArt, the semantization road environment information overseas broadcast of generation is received for car-mounted terminal, is the automatic Pilot service of vehicle.

The above embodiments are merely illustrative of the technical solutions of the present invention rather than is limited, without departing substantially from essence of the inventionIn the case of refreshing and its essence, those skilled in the art make various corresponding changes and become in accordance with the present inventionShape, but these corresponding changes and deformation should all belong to the protection domain of appended claims of the invention.

Claims (10)

1. it is a kind of based on bus or train route collaboration automatic Pilot context aware systems, it is characterised in that it is described based on bus or train route collaboration fromDynamic driving environment sensory perceptual system, is included in trackside sensor, the trackside processing unit set on the section by setpoint distance divisionWith trackside communication unit, wherein：

The trackside sensor includes camera, laser radar and microwave radar, for gathering the road information of road；

The trackside processing unit, for receiving the road information of trackside sensor collection, system is transformed into after processing is identifiedIn one world coordinate system, then transformed data fusion is labeled in high-precision map, to adjacent segments accuratelyFigure forms semantization road environment information after being spliced, and is transmitted to trackside communication unit；

The trackside communication unit, for being sent semantization road environment information by wireless communication technique.

2. the automatic Pilot context aware systems as claimed in claim 1 based on bus or train route collaboration, it is characterised in that the shootingHead collection section video image, identifies kind of object in video image, shape, positional information.

3. the automatic Pilot context aware systems as claimed in claim 2 based on bus or train route collaboration, it is characterised in that the tracksideProcessing unit is receiving the road information of trackside sensor collection, when being transformed into after being identified in unified world coordinate system, holdsThe following operation of row：

All trackside sensors are added to unified world coordinate system；

The kind of object of camera acquisition, shape, positional information are transformed into unified world coordinate system；

The laser point cloud being reflected back that laser radar obtains is received, three-dimensionalreconstruction is carried out to surrounding enviroment, is transformed into the unified worldIn coordinate system；

The microwave being reflected back that microwave radar obtains is received, identification has clear and corresponding position, is transformed into the unified worldIn coordinate system.

4. the automatic Pilot context aware systems as claimed in claim 1 based on bus or train route collaboration, it is characterised in that the tracksideProcessing unit is receiving the road information of trackside sensor collection, when being transformed into after being identified in unified world coordinate system, holdsThe following operation of row：

All trackside sensors are added to unified world coordinate system；

The video image of camera collection is received, identifies kind of object in video image, shape, position, is transformed into unificationIn world coordinate system；

The laser point cloud being reflected back that laser radar obtains is received, three-dimensionalreconstruction is carried out to surrounding enviroment, is transformed into the unified worldIn coordinate system；

The microwave being reflected back that microwave radar obtains is received, identification has clear and corresponding position, is transformed into the unified worldIn coordinate system.

5. the automatic Pilot context aware systems as claimed in claim 1 based on bus or train route collaboration, it is characterised in that the tracksideSensor, further includes weather detector.

6. the automatic Pilot context aware systems as claimed in claim 1 based on bus or train route collaboration, it is characterised in that the tracksideCommunication unit is additionally operable to receive the road status information that administrative center issues, and is sent to trackside processing unit；

The trackside processing unit, is additionally operable to receive the road status information that trackside communication unit is sent, is labeled in accuratelyIn figure.

7. the automatic Pilot context aware systems as claimed in claim 6 based on bus or train route collaboration, it is characterised in that described high-precisionSpending map includes four layers, and first layer is the static high-precision cartographic information of default road, and the second layer is road status information layer, theThree layers are road dynamic environment Information Level, and the 4th layer is road traffic state Information Level.

8. the automatic Pilot context aware systems as claimed in claim 7 based on bus or train route collaboration, it is characterised in that described to turnData fusion after changing is labeled in high-precision map, including：

The road status information received by trackside communication unit is labeled in the second layer；

The road information obtained by trackside sensor is labeled in third layer；

According to the road status information received by trackside communication unit and the road information obtained by trackside sensor, fusionMacro-traffic flow condition information is generated, is labeled in the 4th layer.

9. the automatic Pilot context aware systems as claimed in claim 1 based on bus or train route collaboration, it is characterised in that the tracksideProcessing unit forms semantization road environment information after splicing to the high-precision map of adjacent segments, performs following operation：

The high-precision map of front and rear adjacent segments is spliced respectively, generates the map of two splicings, then two are splicedMap forms semantization road environment information.

10. the automatic Pilot context aware systems as claimed in claim 1 based on bus or train route collaboration, it is characterised in that described to take the photographAs head and microwave radar are laid in above road on rod piece, the laser radar is laid on the rod piece of road center dividing strip,The trackside processing unit, trackside communication unit are located on the rod piece of trackside.