KR20180052811A - Method for Making Autonomous or Automated Driving System and a Driving System - Google Patents

Abstract

The present embodiment relates to an autonomous driving system. More particularly, the present invention relates to a method of making an autonomous driving system by learning from traveling experience data including the traveling information of road users, and to such an autonomous driving system. The method includes: a step of accumulating driving experience data; and a step of completing an autonomous driving algorithm.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention [0002] The present invention relates to an autonomous driving system,

This embodiment relates to an autonomous navigation system. Particularly, the present invention relates to a method for manufacturing an autonomous traveling system by learning from traveling experience data including traveling information of road users, and to such an autonomous traveling system.

The contents described below merely provide background information related to the present embodiment and do not constitute the prior art.

Recently, research on autonomous driving has been actively carried out. For autonomous driving, it is necessary to determine the driving conditions such as driving direction and speed based on the accurate recognition of the external environment through the sensors and the recognized information.

Although radar is used as a sensor for external environment recognition, the use of a vision sensor is becoming more active in order to recognize more information. Vision sensors are also popular because they are relatively inexpensive compared to other sensors. In this regard, the recognition technology of the outside environment by the pattern recognition and the image processing is greatly developed and it is expected to greatly contribute to the autonomous driving.

On the other hand, in order to build an intelligent transportation system, many countries have been attracting a lot of attention from many countries, and related international standards are being prepared. For example, ISO / TC 204 has established standards in relation to the Message on the Road Guidance Protocol (RGP) and Unified Gateway Protocol (UGP), and the Local Dynamic Map (LDM) in ETSI, CEN / TC278, , 'Cooperative Awareness Messages (CAMs)' and 'Decentralized Environmental Notification Messages (DENMs)'.

In particular, the LDM can be classified into four types, from Type 1 to Type 4, according to the dynamic characteristics of information related to the map information. Type 2 information is' quasi-staic 'information such as landmark, traffic sign, etc., Type 3 information is' Dynamic', and ' 'Type 4 is' Highly Dynamic' information, which is information about nearby vehicles, pedestrians, etc. This information corresponds to traffic congestion, traffic light information, traffic accident information, construction section information, and road surface information. Type 1 information can be a very dynamic piece of information that changes over a period of a few seconds if Type 1 information is dynamic information that varies over months or years.

Although LDM is very important for intelligent transportation systems, it is recognized that it is necessary to deal with more precise information in order to be used for autonomous driving. For example, only the Type 1 information should be the level of the three-dimensional map data, not the level of the existing two-dimensional map data. In other words, it is widely recognized that a high-precision three-dimensional map is required for autonomous driving. We are investing huge amounts of capital in Google, Uber, here and others to develop such maps. It is expected to be.

As 3-dimensional high-precision maps are developed and the accuracy of recognition of the surrounding situation by sensor is higher, autonomous driving technology is expected to greatly develop. However, There is still a lack of discussion about it. If the three-dimensional map and the sensor are eyes in the autonomous vehicle, it is necessary to further discuss the brain for the autonomous driving judgment. That is, it is necessary to develop a running program that will perform autonomous driving using digital map information and sensor information.

Even if such a program is developed, a simple program using only the road situation information received from the intelligent traffic system (ITS) as it is currently being discussed is not sufficient to complete autonomous driving.

It is necessary not only to comprehensively consider the weather information, the road surface state information of the road, the traffic road information of the driving road section and the surrounding sections, and the like, including the real-time driving information of the surrounding vehicles, .

Particularly, since the driving experience of humans is an experience of performing safe driving at that time by comprehensively considering the external environment such as the weather or road surface condition and the driving condition of the surrounding vehicles, if such experience data can be utilized in the autonomous driving program, It will be useful.

Combining such experience data with artificial intelligence (AI), which is being actively researched recently, a very useful autonomous driving program can be created.

The present invention is intended to supplement the above-mentioned deficiencies with respect to autonomous driving.

We propose a method of completing an autonomous driving algorithm (or system) by learning autonomous driving system using driving experience data of drivers, and propose an autonomous driving system that performs autonomous driving using the learned autonomous driving algorithm.

One embodiment of the autonomous traveling system manufacturing method includes accumulating a plurality of traveling experience data and learning and completing an autonomous traveling algorithm using the accumulated traveling experience data.

The step of accumulating the plurality of traveling experience data is performed by obtaining the traveling information of the first load user and the traveling information of the peripheral load user of the first load user as the event set for the first load user occurs.

Here, the event may be an event such as sudden braking, sudden acceleration, rapid deceleration, rapid acceleration, lane change, steering angle change, airbag deployment, collision or collision accident, Accident, etc.).

In addition, the event may be a satisfying condition of a specific condition set by the system. For example, when the traveling speed or the acceleration is exceeded or less than the set traveling speed, the lane change, acceleration / deceleration more than the set number of times during the set time, and the like can be set as an event. The event may be determined differently depending on the content of the driving experience data to be obtained.

The running information of the peripheral load user or the running information of the first load user may include the running information at the time of occurrence of the event or the running information before or after the occurrence time of the event.

The learning stage can be performed through artificial intelligence algorithms. The running information of the first load user before the occurrence of the event and the running information of the peripheral load user are used as input data and the steering angle information, the braking information, the acceleration information, the deceleration information, the transmission information, The autonomous traveling algorithm can be completed through learning by using at least one of the information as output data.

On the other hand, the road user includes a person or an object temporarily occupying and using the road. For example, it may be a temporary obstacle such as a pedestrian, a bicycle with a person, a rockfall, a four-wheeled vehicle such as a two-wheeled vehicle such as a motorcycle, a passenger car, a truck or a bus.

The peripheral load user may be a load user existing within the setting range from the first load user.

The setting range may be set to include only the rectilinear load user of the first load user, but it may be set to a wider range, so that the related peripheral load users are further increased.

For example, the peripheral load user includes at least one of a left front, an electrostatic room, a right front, a left room, a right room, a left rear, a front right, and a rear right of the first load user.

The peripheral load user may be able to sense the travel information himself or receive it from the outside. If the surrounding load user is falling, this corresponds to an obstacle on the road, and the obstacle may also be the surrounding load user. However, since the information can not be obtained separately from the surrounding load user, the other load user or the first load user .

The driving information may include at least one of the type of the load user, the traveling direction, the traveling speed, the type of traveling road, the driving lane, GPS information, the braking information, and the steering angle information.

The traveling experience data may further include at least one of climatic information of a region where the first load user travels, road surface state information of the road, traffic congestion information, traffic accident information, construction section information, and road obstacle information. Thus, autonomous algorithms can be learned using such information together.

Meanwhile, an embodiment of the autonomous driving system according to the present invention is an autonomous driving system produced by the above-described method.

In addition, the autonomous navigation system can be mounted on the vehicle and can be used to receive GPS information of a vehicle, set a route to a destination, start autonomous travel along a set route, The method comprising the steps of sensing information by a road vehicle, continuing autonomous travel along a set route by using GPS information and lane information, acquiring travel information of a peripheral load user, learning by artificial intelligence algorithm by the above- And changing the running state of the subject vehicle according to the running information of the peripheral load user using the autonomous running algorithm.

In carrying out the autonomous driving carried out on the vehicle, in addition to the traveling information of the surrounding road users, information on the climate of the area where the vehicle is traveling, road surface state information, traffic congestion information, traffic accident information, construction section information, , And this information may be received from a server of a specific organization that handles the information, or may be transmitted from a roadside base station in the corresponding area.

It is possible to obtain an autonomous running system capable of realizing safe operation in real time in response to the running state of the surrounding vehicles.

An autonomous driving system that can cope with safe driving in various situations can be obtained by learning through artificial intelligence algorithm using human driving experience data.

Further, since the human experience data is used, an autonomous run very similar to a human running habit can be achieved, and therefore, an autonomous running system that performs familiar running without discomfort can be obtained.

1 is a view schematically showing a state in which traveling information is acquired from a first load user and load users around the first load user to secure traveling experience data.
Fig. 2 is a flowchart for explaining the process of obtaining travel experience data.
FIG. 3 is a diagram schematically showing how an autonomous driving algorithm is learned using traveling experience data. FIG.
4 is a view schematically showing a state in which the vehicle is autonomously driven using an autonomous vehicle on which the autonomous vehicle is mounted.

Hereinafter, embodiments of the present invention will be described in detail with reference to exemplary drawings. It should be noted that, in adding reference numerals to the constituent elements of the drawings, the same constituent elements are denoted by the same reference symbols as possible even if they are shown in different drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

In describing the components of the present invention, terms such as first, second, A, B, (a), and (b) may be used. These terms are intended to distinguish the constituent elements from other constituent elements, and the terms do not limit the nature, order or order of the constituent elements. Throughout the specification, when an element is referred to as being "comprising" or "comprising", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise . In addition, '... Quot ;, " module ", and " module " refer to a unit that processes at least one function or operation, and may be implemented by hardware or software or a combination of hardware and software.

First, a method for acquiring travel information from the first load user and the load users around the first load user to secure travel experience data will be described with reference to FIG.

The first vehicle 210 includes an event sensing unit 110 and a surrounding vehicle information receiving unit 120. The server 130 includes a traveling information receiving unit 132 and a database building unit 134.

The event detection unit 110 of the first vehicle 210 detects whether or not an event set for the first vehicle 210 has occurred.

Here, the event may be an event such as sudden braking, sudden acceleration, rapid deceleration, rapid acceleration, lane change, steering angle change, airbag deployment, collision or collision accident, Accident, etc.).

In addition, an event may be a satisfying condition of a specific condition set by a system. For example, when the traveling speed or the acceleration is exceeded or less than the set traveling speed, the lane change, acceleration / deceleration more than the set number of times during the set time, and the like can be set as an event. The event can be determined differently depending on the contents of the traveling experience data to be obtained.

The traveling information includes at least one of the type of the vehicle, the traveling direction, the traveling speed, the driving lane, the GPS information, the braking information, and the steering angle information.

The neighboring vehicle information receiving unit 122 receives driving information for the neighboring vehicles 220, N1, N2, N3, N4, N5, N6, N7, and 220 of the first vehicle 210. Here, the peripheral vehicle 220 is within the set range from the first vehicle 210, and the peripheral vehicle 220 is located in the left front, the electrostatic room, the right front, the left room, the right room, And the right rear side. Further, the peripheral vehicle 220 is a four-wheeled vehicle, a three-wheeled vehicle, or a two-wheeled vehicle. The surrounding vehicle running information includes at least one of the type of the vehicle, the running direction, the running speed, the type of running road, the driving lane, the GPS information, the braking information, and the steering angle information.

The traveling information receiving unit 132 is included in the server 130 and receives the traveling information of the neighboring vehicle 220 from the neighboring vehicle information receiving unit 122. [ On the other hand, the traveling information receiving unit 132 may receive the traveling information from each of the plurality of neighboring vehicles 220 as shown in FIG.

In addition, the travel information receiver 132 receives the event information and the travel information of the first vehicle 210 from the neighboring vehicle information receiver 122. Here, the running information of the first vehicle 210 includes at least one of the type, the running direction, the running speed, the running lane, the GPS information, the braking information, and the steering angle information of the corresponding vehicle.

Here, the running information of the nearby vehicle or the running information of the first vehicle includes at least one of running information before the occurrence of the event, running information at the time of occurrence of the event, and running information after the occurrence of the event.

Also, the travel information receiver 132 may transmit at least one of the climate information of the area where the first vehicle 210 travels, road surface state information of the road, traffic congestion information, traffic accident information, construction section information, Receive. The transmission of such data is transmitted from an agency or a roadside base station that handles the data.

The database construction unit 134 constructs a database by associating the transmitted neighboring vehicle driving information and the first vehicle driving information with the information about the event.

The network (not shown) refers to a communication network that transmits and receives data using a communication protocol using wired / wireless communication technology. The network transmits and receives data of the event detection unit 110, the neighboring vehicle information receiving unit 120, and the server 130.

The process of securing the running experience data through FIG. 2 will be described in more detail.

First, the travel information of the neighboring vehicle 220 or the travel information of the first vehicle 210 is received (S410).

The reception of the driving information may be received by the first vehicle 210 and transmitted to the server 130 or may be directly transmitted from the neighboring vehicle 220 to the server 130 in the case of the traveling information of the neighboring vehicle 220. [

It is detected whether an event set for the first vehicle 210 has occurred (S420). For example, when the first vehicle 210 suddenly brakes, it detects that an event has occurred.

The detection of the occurrence of such an event may be made in the system in the first vehicle 210, but is not limited thereto.

(At step S430) at least one of climatic information of the area where the first vehicle 210 travels, road surface state information of the road, traffic congestion information, traffic accident information, construction section information, and road obstacle information.

If the event does not occur, the process returns to step S410 in which the driving information of the neighboring vehicle 220 or the driving information of the first vehicle 210 is received until the occurrence of the event (S440). For example, the first vehicle 210 continuously receives driving information from the neighboring vehicles 220, and continuously checks whether an event occurs.

When the event occurs, the server 130 receives the driving information for the neighboring vehicle 220 of the first vehicle 210 (S450). The surrounding vehicle 220 driving information may be transmitted to the first vehicle 210 and transmitted from the first vehicle 210 to the server 130 or may be transmitted directly from the peripheral vehicle 220 to the server 130 have.

When the traveling information of the neighboring vehicle 220 is directly transmitted from the neighboring vehicle 220 to the server 130, for example, the server 130 continuously transmits the traveling information of the first vehicle 210 and / It is possible to secure data by storing driving information of the neighboring vehicle 220 in a memory as a signal indicating that an event has occurred from the first vehicle 210 is received.

In addition, the server 130 receives the event information and the driving information of the direct transmission from the first vehicle-proximate vehicle 220 to the server 130 (S460).

In step S470, the transmitted vehicle navigation information and the running information of the first vehicle 210 are stored in association with the event information in step S470.

Although it is described in FIG. 2 that steps S410 to S470 are sequentially performed, it is only described as an example of the technical idea of the present embodiment. If a person skilled in the art is familiar with the present invention, It will be understood that various modifications and changes may be made to the present invention without departing from the essential characteristics thereof, such as by changing the order described in FIG. 2 or by executing one or more of steps S410 through S470 in parallel, But is not limited thereto.

The traveling experience data constructed by the above-described method is used for learning an autonomous driving program or system, as shown in FIG.

In other words, by using the surrounding vehicle driving information and the first vehicle driving information of various variables occurring in various situations as learning data, the autonomous driving algorithm is learned through the artificial intelligence algorithm.

The steering angle information, the braking information, the acceleration information, and the steering angle information are used as the running behavior of the first vehicle 210 at the time of occurrence of the event, with the running information of the first vehicle 210 and the running information of the nearby vehicle 220 before the occurrence of the event, Deceleration information, transmission information, and engine fuel supply information as output data to learn the autonomous driving algorithm through deep learning.

At this time, at least one of the climate information of the area where the first vehicle 210 travels when the event occurs, road surface state information of the road, traffic congestion information, traffic accident information, construction section information, As a learning material. Such travel environment information such as climate information can be obtained from an agency handling the data or a roadside base station server.

The process of autonomous traveling performed by the autonomous traveling system created by the method described above with reference to FIG. 4 will be described.

First, the autonomous traveling system is installed in the first vehicle 210 as a host vehicle.

Three cars are running on the front side of the first vehicle 210 and a two-wheeled vehicle 214 is on the left side and a four-wheeled vehicle 215 And two two-wheeled vehicles 216 and 217 and one two-wheeled vehicle 218 are running on the rear side.

The first vehicle 210 is in a V2V communication state with the nearby vehicles 211 to 218 and receives the traveling information of the nearby vehicles 211 to 218 through communication. In addition, some or all of the traveling information of the nearby vehicles 211 to 218 may be secured through a sensor (not shown) mounted on the first vehicle 210 in addition to reception through communication.

The first vehicle 210 is in communication with the environment information server 300. The first vehicle 210 receives environment information from the environment information server 300 as climate information of the traveling region, road surface state information of the road, traffic congestion information, Construction section information, and obstacle information on the road.

On the other hand, forward traffic congestion information, traffic accident information, construction section information, road obstacle information, and the like may be received not only from the environment information server 300 but also from other vehicles traveling ahead through V2V. It is possible to overcome the limitations of the sensor through the information exchange by the V2V communication with the other vehicle and to acquire the information in more real time than the case through the server 300. [

The autonomous vehicle traveling system mounted on the first vehicle 210 outputs data for driving control of the first vehicle 210 using the running information and / or running environment information of the nearby vehicles 211 to 218 as input data. For example, at least one of steering angle data, braking data, acceleration data, deceleration data, transmission data, and engine fuel supply data is output as running behavior control data.

Then, the data thus outputted is used as control data of the system or part so that the running of the first vehicle 210 is controlled. For example, the steering angle data output as the driving behavior control data is used to control the steering of the first vehicle 210, and the braking data is used to control the braking of the first vehicle 210. [

The autonomous traveling system performs autonomous traveling with respect to the first vehicle 210 while repeating this process.

Since the autonomous driving system includes an algorithm learned through deep running or the like from human driving experience data for achieving safe driving with respect to driving information and driving environment information of nearby vehicles, It is possible not only to make coping possible, but also to achieve a driving behavior similar to a human driving habit, thereby achieving an autonomous driving without discomfort.

In the above-described embodiment, the load user is a vehicle, but the present invention is not limited thereto.

It is to be understood that the above description is only illustrative of the technical idea of the present embodiment and that various modifications and changes may be made by those skilled in the art without departing from the essential characteristics of the embodiments . Therefore, the present embodiments are to be construed as illustrative rather than limiting, and the scope of the technical idea of the present embodiment is not limited by these embodiments. It is to be understood that the scope of the present invention is to be construed as being limited only by the scope of the appended claims.

110: Event detection unit
120: nearby vehicle information receiver
130: Server
132:
134: Database construction part
210: a first vehicle (or a child vehicle)
211: First peripheral vehicle
212: second surrounding vehicle
213: Third neighboring vehicle
214: fourth peripheral vehicle
215: the fifth neighboring vehicle
216: 6th peripheral vehicle
217: Seventh neighboring vehicle
218: the eighth peripheral vehicle
300: travel environment information server

Claims (13)

Accumulating the traveling information of the first load user and the traveling information of the peripheral load user of the first load user in accordance with the occurrence of the event set for the first load user and accumulating the plurality of traveling experience data; And
Completing an autonomous traveling algorithm through learning using the plurality of traveling experience data
Wherein the method comprises the steps of:
The method according to claim 1,
Wherein the event is at least one of sudden braking, sudden acceleration, rapid deceleration, sudden acceleration, lane change, steering angle change, airbag deployment, crash or collision,
Alternatively, the event is a satisfaction of a systematically set specific condition of the first load user.
The method according to claim 1,
Wherein the running information of the peripheral load user or the running information of the first load user includes at least one of the running information before the occurrence of the event and the running information after the occurrence of the event.
The method according to claim 1,
Wherein the learning step is performed through an artificial intelligence algorithm.
5. The method of claim 4,
Wherein the learning step sets, as input data, the running information of the first load user before the occurrence of the event and the running information of the neighboring load users as input data, and calculates steering angle information, braking information, Wherein at least one of information, deceleration information, transmission information, and engine fuel supply information is used as output data to learn the artificial intelligence algorithm.
The method according to claim 1,
Wherein the peripheral load user is within a setting range from the first load user.
The method according to claim 1,
Wherein the peripheral load user includes at least one of a left front, an electrostatic front, a right front, a left room, a right room, a left rear, a forward rear, and a rear right with respect to the first load user How to make a traveling system.
The method according to claim 1,
The peripheral load user can transmit his / her travel information to another system via wireless communication, can not obtain his / her own travel information, can not transmit his / her travel information to another system via wireless communication Or at least one of < RTI ID = 0.0 >
Or at least one of a pedestrian, a bicycle, a four-wheeled vehicle, a three-wheeled vehicle, or a two-wheeled vehicle.
The method according to claim 1,
Wherein the first load user travel information or the surrounding road user travel information includes at least one of a type of the vehicle, a traveling direction, a traveling speed, a type of traveling road, a driving lane, GPS information, braking information, A method of manufacturing an autonomous traveling system.
The method according to claim 1,
The traveling experience data may further include at least one of climate information of a region where the first load user travels, road surface state information of a road, traffic congestion information, traffic accident information, construction section information, Wherein said method comprises the steps of:
An autonomous traveling system produced by any one of claims 1 to 10.
Receiving GPS information of the vehicle;
Setting a path to a destination;
Starting autonomous travel along the set route;
Sensing information by a vehicle on the road on which the vehicle is traveling;
Continuing autonomous travel along the set route using the GPS information and lane information;
Acquiring travel information of a peripheral load user;
Accumulating the traveling information of the first load user and the traveling information of the peripheral load users of the first load user in accordance with the occurrence of the event set for the first load user to accumulate a plurality of traveling experience data, A step of changing a running state of the subject vehicle according to running information of the peripheral load user using an autonomous driving algorithm learned through an artificial intelligence algorithm using data
And an autonomous drive system.
13. The method of claim 12,
The step of changing the traveling state of the subject vehicle further includes, in addition to the traveling information of the user of the surrounding road, the weather information of the area where the first load user travels, the road surface state information of the road, the traffic congestion information, Information on the road, and obstacle information on the road.

Images