KR101871441B1 - Voice recognition field survey and road linear generation system for high definition electronic map production for autonomous vehicle - Google Patents

Abstract

The present invention relates to a voice recognition field survey and road linear generation system for the high definition electronic map production for an autonomous vehicle. More specifically, the voice recognition field survey and road linear generation system for the high definition electronic map production for an autonomous vehicle allows an automatic production of road lines when producing a high definition electronic map for an autonomous vehicle using a field investigation method by utilizing voice recognition, image information obtained by photographing the surrounding environment around the road, and positioning information.

Description

[0001] The present invention relates to a voice recognition field survey and road linear generation system for producing a high-precision electronic map for an autonomous vehicle,

The present invention relates to a voice recognition field survey and a road line generation system for producing a high-precision electronic map for an autonomous vehicle, and more particularly, The present invention relates to a voice recognition field survey and a road linear generation system for producing a high-precision electronic map for an autonomous driving vehicle, which can automatically generate a road alignment when a high-precision electronic map for a driving vehicle is produced.

The current on-site survey system for the production of high-precision electronic maps is equipped with DGPS (Differential GPS) receiver and many cameras and equipment such as IMU (inertial measuring unit) and LiDAR (Light Detection and Ranging) It is configured to scan roads and roadside facilities through a system called MMS (Mobile Mapping System).

Globally, these instruments are used to improve the accuracy of map field investigations from tens of meters to 50 cm.

On the other hand, autonomous vehicles should be equipped with state-of-the-art technologies that operate the vehicle by themselves without driver intervention, and these technologies are realized through numerous sensors and communication technologies.

However, it is essential that autonomous vehicles have precise electronic maps, including key information on roads, in case the malfunctions or obstacles of the configured devices do not lead to normal operation.

However, existing electronic maps with tens of meters of error can not be used due to their low accuracy.

Therefore, in order to ensure the accuracy of positioning error less than 50 cm, high-precision field positioning is performed using a map field survey system equipped with MMS equipment.

As described above, the MMS equipment consists of four (DGPS, IMU, camera, LiDAR) core equipment and performs high precision positioning.

At this time, the GPS uses DGPS, which is precisely positioned, to dramatically improve the absolute position accuracy from several tens of meters to 50 cm of the error of the existing GPS.

In addition, LiDAR is a relative positioning method that measures the distance of the target point based on the absolute position of all the facilities around the road. The target point can be expressed as a point and can be displayed as 700,000 to 1 million points per second, Most of the facilities in the vicinity of the road can be represented by a point cloud. The measurable range is generally within 70m and the accuracy is less than 2cm.

In addition, the IMU continuously measures the relative distances such as height, curvature, etc. based on speed, acceleration, etc., and calculates the DGPS .

In addition, the camera usually mounts 4 ~ 6 cameras, and records 360, 360, and 360 frames in one frame at the same time.

Currently, high-precision electronic maps are being produced using data acquired, recorded, and stored through such high-quality equipment.

However, the core of high-precision electronic maps is that all lanes should be linearly shaped for precise autonomous driving.

However, the existing electronic map is composed of one road linear shape, or two road linear shapes such as wide road and highway for map matching.

Here, the most important point of the high-precision electronic map is that a road network should be produced based on all lanes.

Therefore, the existing production method is to use the received GPS data to create a center line and input all the road information to the corresponding center line. However, the high precision electronic map production method uses the precisely measured DGPS to travel through the image If you use lane information, you have to create 4 lines of linearity at regular intervals in 4 lanes and enter each property.

Therefore, since it is necessary to produce several road lines based on the lane identified through the photographed image, it takes more than five times as long as the conventional electronic map production time, and as a result, it becomes difficult to update quickly.

As such, the electronic map makers in Korea are very passive in the production of high-precision electronic maps, and currently there is no way to improve the investment cost and the input time.

Korean Patent Registration No. 10-0508974 (Aug. 2005) 'Electronic map construction system using road information and its method'

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems of the prior art, and it is an object of the present invention to provide a high-precision electronic map for an autonomous vehicle using a field survey method using voice recognition, The present invention provides a voice recognition field survey and a road linear generation system for producing a high-precision electronic map for an autonomous driving vehicle capable of automatically generating a road alignment.

A DGPS receiving unit (102) for receiving position information from a GPS satellite (100) and precisely correcting the position information; An audio input unit 104 for inputting and recognizing the voice of the researcher who is surveyed in the field through the microphone 106; The images generated by the first camera 122 and the second camera 124, the third camera 126 and the fourth camera 128 installed on the front, rear, left, and right sides of the vehicle are stored An LiDAR DATA positioning and storage unit 116 for recognizing and storing the facility type around the road, an IMU DATA positioning and storing unit 118 for positioning and storing the speed, slope, and height values of the vehicle, A received data storage and processing unit 112 including a user data processing unit 120 for providing and processing an interface (UI) with a user and storing and processing the received data provided by them; A display unit 108 for displaying data and images provided by the received data storage and processing unit 112 and positional information received from the DGPS receiving unit 102; The current position coordinates are calculated using the position information from the DGPS receiving unit 102, and the map data of the currently photographed point is extracted from the memory unit 148, and the camera image of the photographing point, the LiDAR data, and the IMU data The memory unit 148 includes an electronic map data storage DB 132 for storing electronic map data, a user data storage DB 132 for storing information input through a user interface, An image data storage DB 136 for storing image information received from the first, second, third and fourth cameras 122, 124, 126 and 128; an audio data storage DB 136 for storing audio information input through the audio input unit 104; A LiDAR data storage DB 140 for storing LiDAR information provided from the LiDAR DATA positioning and storage unit 116, a new road information storage unit 140 for storing new road information provided from the national geographical information source, A DGPS data storing DB 142 for storing DGPS information provided from the DGPS receiving unit 102, an IMU data storing DB 142 for storing positioning information provided from the IMU DATA positioning and storing unit 118, (144); A position where the absolute position positioned through the DGPS receiving unit 102 and the absolute position corrected through the IMU on the opponent side in a place where the DGPS including the tunnel is not received is set as the position information of the vehicle; The distance to the right boundary is used to measure the relative distance between the vehicle and the facility located through LiDAR and the distance to the left boundary to the relative distance between the vehicle and the facility located through LiDAR; (Gx-Lx |) + (| Gx-Rx |), the road width (X) is first calculated in terms of units of m, , And a road line shape is automatically generated on the left and right sides of the driving lane of the vehicle provided by the input voice information calculated as the lane width = X / total total line. And a road linear generation system.

Here, Gx is the X coordinate value of the vehicle position information, Lx is the X coordinate value of the distance from the left boundary stone, and Rx is the X coordinate value of the distance from the right boundary stone.

According to the present invention, road linearity can be automatically generated when a high-precision electronic map for an autonomous vehicle is produced using a field survey method using voice recognition, image information obtained by photographing the road periphery, and positioning information.

FIG. 1 is an exemplary block diagram of a voice recognition field survey and a road alignment system for producing a high-precision electronic map for an autonomous vehicle according to the present invention.
FIG. 2 is an exemplary flow chart illustrating a field recognition process for producing a high-precision electronic map for an autonomous vehicle according to the present invention, and a field survey process using a road line generation system.
FIG. 3 is an exemplary plan view illustrating an example in which a road line by car line is automatically generated through speech recognition in the system according to the present invention.
4 is a conceptual diagram for explaining the principle of road linear automatic generation according to the present invention.
FIG. 5 is a view showing an example of a dust and damping structure of a control box and a DB case installed in a vehicle constituting the system according to the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Before describing the present invention, the following specific structural or functional descriptions are merely illustrative for the purpose of describing an embodiment according to the concept of the present invention, and embodiments according to the concept of the present invention may be embodied in various forms, And should not be construed as limited to the embodiments described herein.

In addition, since the embodiments according to the concept of the present invention can make various changes and have various forms, specific embodiments are illustrated in the drawings and described in detail herein. However, it should be understood that the embodiments according to the concept of the present invention are not intended to limit the present invention to specific modes of operation, but include all modifications, equivalents and alternatives falling within the spirit and scope of the present invention.

As shown in FIG. 1, a voice recognition field survey and a road linearization system for producing a high-precision electronic map for an autonomous vehicle according to the present invention includes a DGPS receiver 102 )Wow; An audio input unit 104 for inputting and recognizing the voice of the researcher who is surveyed in the field through the microphone 106; The images generated by the first camera 122 and the second camera 124, the third camera 126 and the fourth camera 128 installed on the front, rear, left, and right sides of the vehicle are stored An LiDAR DATA positioning and storage unit 116 for recognizing and storing the facility type around the road, an IMU DATA positioning and storing unit 118 for positioning and storing the speed, slope, and height values of the vehicle, A received data storage and processing unit 112 including a user data processing unit 120 for providing and processing an interface (UI) with a user and storing and processing the received data provided by them; A display unit 108 for displaying data and images provided by the received data storage and processing unit 112 and positional information received from the DGPS receiving unit 102; The current position coordinates are calculated using the position information from the DGPS receiving unit 102, and the map data of the currently photographed point is extracted from the memory unit 148, and the camera image of the photographing point, the LiDAR data, and the IMU data The memory unit 148 includes an electronic map data storage DB 132 for storing electronic map data, a user data storage DB 132 for storing information input through a user interface, An image data storage DB 136 for storing image information received from the first, second, third and fourth cameras 122, 124, 126 and 128; an audio data storage DB 136 for storing audio information input through the audio input unit 104; A LiDAR data storage DB 140 for storing LiDAR information provided from the LiDAR DATA positioning and storage unit 116, a new road information storage unit 140 for storing new road information provided from the national geographical information source, A DGPS data storing DB 142 for storing DGPS information provided from the DGPS receiving unit 102, an IMU data storing DB 142 for storing positioning information provided from the IMU DATA positioning and storing unit 118, (144).

The system according to the present invention having such a configuration is utilized for field investigation in the form of FIG. 2 and FIG.

First, when the field survey is started, the field survey system according to the present invention is booted.

Then, the mode is switched to the field survey mode.

At this time, if the field survey mode is On, it starts driving. Otherwise, it confirms whether to continue the investigation. If the investigation is finished, the process ends. If the investigation is continued, the field survey mode is kept on.

If it is determined that the driving is to be continued and the driving is to be terminated, the operation is fed back to the field survey mode switching step. If the driving is continued, it is checked whether various signals are normally received. If the signals are normally received, If the current location is confirmed, the map is loaded and the DGPS signal received according to the current location is displayed on the display unit.

At this time, when a predetermined voice signal is received, a change point (Node) is generated at the corresponding position.

If it is determined that the predetermined voice signal is not correct, the voice signal is requested to be input again.

When the information is inputted through voice recognition, the input contents are displayed on the screen on the basis of the inputted information. The process of determining whether the information is a first lane or a second lane after the change point shown in FIG. 2 is generated, And a process of storing and displaying a line generated from the step of determining whether the line is displayed.

However, since this process is performed in a time-wise manner, a detailed explanation is omitted.

In this process, the voice signal is repeatedly received until the voice signal of 'end' is received, and when the voice signal of 'end' is received, the field investigation process itself is terminated.

At this time, speech recognition is performed according to a predetermined rule, which can be modified at any time since it is done by defining a rule.

For example, if a voice of 'primary one in five lanes' comes in, one lane five lanes are drawn on the screen based on the position, and the road linear of the primary lane during the current driving is displayed in red.

This means that the road line type road line type is automatically generated through voice recognition, and the principle of road line type automatic generation will be described later with reference to FIG.

This can be used when the opposite lane is not identified (boundary / median etc.), or when the lane on the opposite lane is different from the lane on the other side, or especially when collecting information on one way.

As another example, when a voice called '5-car primary in 9 round trip rounds' is received, it corresponds to the case of acquiring information by round trip. It means 9 lanes in total, and the present position is based on the leftmost lane 5th, and means that the vehicle is traveling in the primary direction in the running direction, and in this case, the road line form of the running lane can be displayed in red as in the one way. This indicates that the center line is between the fourth lane and the fifth lane with respect to the leftmost lane of the road.

The collected information is used to automatically generate the road line shape, and it is generated as a linear network for the high precision electronic map, which is not a simple linear line formed in the past.

In other words, in the past, a road line shape was selected by looking at a photographed image. However, it is possible to automatically produce a road line shape by car line based on the voice obtained from the field. When the road line shape is obtained in this manner, And the production time can be drastically reduced.

The principle of road linear autogeneration will now be described with reference to FIG.

Referring to FIG. 4, the absolute position of the vehicle is applied through the DGPS receiver, and the absolute position of the vehicle is corrected through the IMU when the DGPS is not received, such as in a tunnel.

In addition, the distance to the right side boundary is used to measure the relative distance between the vehicle and the facility (shape), which is located through LiDAR, and the distance to the left side boundary is the same.

In addition, the car player applies the car entered through voice information, and the unit is converted into m.

Based on these assumptions, first calculate the road width.

The road width is calculated as X = (| Gx-Lx |) + (| Gx-Rx |).

4, the vehicle position information is represented by (Gx, Gy), the distance from the right side boundary is expressed by (DELTA Rx, y), and the distance from the left side boundary is indicated by (DELTA Lx, The output uses these values.

For example, if the road width X is 19.5 m and the voice information is entered as 'third-order second order in six round-trip lanes'

Lane width = X (19.5m) / 6 (total difference) = 3.25m.

Therefore, in the road linear automatic generation, one road linear form is generated based on the survey vehicle, one road is formed at 3.25m intervals on the right side, and four road linear forms are arranged on the left side at 3.25m intervals.

In this case, the voice information means a total of six lanes on a round trip basis, the current position is the third from the leftmost lane, the second lane in the running direction, and the center line refers to the leftmost lane of the road And that it is between the first lane and the second lane.

As described above, according to the present invention, it is possible to automatically generate roads linearly on the basis of the above-mentioned system and speech recognition, thereby making it possible to reduce the time required for producing electronic maps, thereby enabling quick updating.

5, the control unit 110 is mounted on the control box 200, and the memory unit 148 includes a DB case 210 fixed adjacent to the control box 200. In addition, A plurality of DBs are mounted in the form of a server rack ship inside. These control box 200 and DB case 210 are fixed to the bottom surface of the vehicle so that they are generated when the vehicle vibrates or the road surface is uneven. Which may be affected by impact or vibration.

Therefore, in order to minimize such influence, the vibration damping sheet 300 is further provided on the bottom surface of the control box 200 and the bottom surface of the DB case 210 to prevent vibration and shock. Electrical stability can also be maintained if it is equipped with insulation.

The vibration damping buffer sheet 300 preferably has a hexagonal honeycomb-like elastic recovery structure and is preferably bonded and fixed with an adhesive.

At this time, the hexagonal honeycomb shape is formed by a combination of polypropylene and PET (polyethylene terephthalate) in order to sufficiently retain elastic restoring force.

For example, the vibration damping buffer sheet 300 may include upper frame filaments 310 and upper frame filaments 320 forming a lower layer and upper frame filaments 310 and 320, respectively, The upper and lower frame filaments 310 and 320 are connected to each other by connecting a plurality of bundling filaments 330 so as to increase the collecting force while increasing the elastic force.

In this case, it is preferable that the upper and lower frame filaments 310 and 320 are made of a coarse yarn, such as a yarn of 60 filament -300 denier.

This is to maintain the frame, and when the thinner yarn is used, the shape is broken and it is difficult to maintain the shape of the vibration damping buffer sheet 300.

Particularly, the bundling filaments 330 are filaments thinner than the upper and lower frame filaments 310 and 320 so as to constrain the upper and lower frame filaments 310 and 320 and to maintain a gap therebetween, That is, it is curled in a curved shape that is distributed to one side to form a collecting space, thereby enhancing noise and dust collecting power, and further strengthening the anti-vibration force by securing the elastic absorbing power.

In addition, the vibration damping buffer sheet 300 may be prepared by mixing a mixture of hematite (Fe ₂ O ₃ ) and chromium oxide (Cr ₂ O ₃ ) in a weight ratio of 1: 1 so as to enhance heat resistance, insulation resistance, chemical resistance and durability By weight of sodium bicarbonate and 1.5-2.5% by weight of borax, calcining the calcined body at a temperature of 800 to 1200 ° C, wet-crushing the calcined body with a crusher to have a particle size of 1 μm or less, 20% by weight of an oxide powder dried at 100 DEG C for 10 hours; 5.5 weight percent of a mixture of germanium and gallium arsenide and zinc selenide in a weight ratio of 1: 1: 1; 3.0% by weight of zinc oxide (Zn (NO ₃ ) ₂ .6H ₂ O); 2.5% by weight of allopene powder; 2.5% by weight of aluminum hydroxide powder; 2.5% by weight of nickel titanate; First aluminum phosphate _{[Al (H 2 PO 4)} 3] and the first of magnesium phosphate _{[Mg (H 2 PO 4)} 2] is 1 to 10% by weight of the ginseng salt mixed in a weight ratio of 1; 10% by weight of xylene and the remaining liquid unsaturated polyester resin is impregnated for 24 hours.

In this case, the hematite and chromium oxide constituting the oxide powder are components that maintain the infrared reflectance of 30% or more with the color development. Sodium bicarbonate causes the vaporized sodium gas to act on the oxide to form a glass phase on the oxide surface, And borax is a component added to enhance the bonding force while smoothing the oil surface of the oxide surface.

When the firing temperature is less than 800 ° C., the coloring degree is drastically lowered and the infrared absorption rate is increased, so that the function of the present invention as an infrared reflective pigment can not be achieved, and the firing temperature of 1200 ° C. The magnetization phenomenon occurs. Therefore, the temperature should be limited to the above-mentioned range and fired.

Particularly, when high-temperature firing is performed, oxides are compounded and synthesized in addition to the above-mentioned characteristics, thereby forming a strong crystal structure and also having chemical characteristics excellent in weather resistance, heat resistance, chemical resistance and safety.

Here, germanium, gallium arsenide, and zinc selenide are materials for enhancing electrical insulation. Especially, when they are mixed in the same ratio, they have excellent insulating properties.

In addition, zinc sulfate is added to enhance heat-resistance stability by forming a eutectic point, and alophene powder is a clay mineral generated in the weathering process of volcanic ash, which is an exceptionally non-crystalline structure, It is added as a powder to enhance the VOC abatement effect.

The aluminum hydroxide powder serves to neutralize the acid, and the powder has a nanometer size. The aluminum hydroxide powder is more preferable because it has a small pore size so that particles and particles are dense and connected to the surface of the dustproof soundproofing sheet This is because the durability is improved as the adhesion is improved.

In addition, the nickel titanate is added to enhance the insulating shielding property while maintaining the stability of the coating liquid.

The phosphate is added to maximize the insulation property of the coating layer, and is added to enhance the film properties of the filament by phosphoric acid, that is, to stabilize the coating layer after coating and to maintain strong bonding force and to suppress film breakdown even under high temperature environment.

100: GPS satellite 102: DPGS receiver
104: audio input unit 108:
110: control unit 112: received data storage and processing unit
148:

Claims (1)

A DGPS receiver (102) for receiving position information from the GPS satellite (100) and precisely correcting the position information; An audio input unit 104 for inputting and recognizing the voice of the researcher who is surveyed in the field through the microphone 106; The images generated by the first camera 122 and the second camera 124, the third camera 126 and the fourth camera 128 installed on the front, rear, left, and right sides of the vehicle are stored An LiDAR DATA positioning and storage unit 116 for recognizing and storing the facility type around the road, an IMU DATA positioning and storing unit 118 for positioning and storing the speed, slope, and height values of the vehicle, A received data storage and processing unit 112 including a user data processing unit 120 for providing and processing an interface (UI) with a user and storing and processing the received data provided by them; A display unit 108 for displaying data and images provided by the received data storage and processing unit 112 and positional information received from the DGPS receiving unit 102; The current position coordinates are calculated using the position information from the DGPS receiving unit 102, and the map data of the currently photographed point is extracted from the memory unit 148, and the camera image of the photographing point, the LiDAR data, and the IMU data And a control unit (110) for performing display control,
The memory unit 148 includes an electronic map data storage DB 132 for storing electronic map data, a user data storage DB 134 for storing information input through a user interface, a first, second, An audio data storage DB 138 for storing audio information input through the audio input unit 104, and an audio data storage DB 138 for storing audio information received from the LiDAR data positioning and storage unit 116 A LiDAR data storage DB 140 for storing LiDAR information provided, a new road data storage DB 146 for storing new road information provided from the terrestrial geographic information source, a DGPS storing DGPS information provided from the DGPS receiving unit 102, A data storage DB 142, and an IMU data storage DB 144 for storing positioning information provided from the IMU DATA positioning and storage unit 118;
A position where the absolute position positioned through the DGPS receiving unit 102 and the absolute position corrected through the IMU on the opponent side in a place where the DGPS including the tunnel is not received is set as the position information of the vehicle; The distance to the right boundary is used to measure the relative distance between the vehicle and the facility located through LiDAR and the distance to the left boundary to the relative distance between the vehicle and the facility located through LiDAR; The car player applies the car entered through the voice information and calculates the road width (X) in units of m,
The road width X = (| Gx-Lx |) + (| Gx-Rx |) is calculated as the lane width = X / And a road linearization system for producing a high-precision electronic map for an autonomous driving vehicle.
(Where Gx is the X coordinate value of the vehicle position information, Lx is the X coordinate value of the distance from the left boundary seat, and Rx is the X coordinate value of the distance from the right boundary seat)

Images