KR102382219B1 - The settlement method and system of expressing FOD on the road for artificial intelligence autonomous driving and the information usage charge by inferring the information value according to the risk of each abnormal object - Google Patents

Abstract

The present invention relates to an autonomous driving service which detects FOD on a road in real time through road condition analysis for artificial intelligence-based autonomous driving and provides information on the detected FOD. An autonomous vehicle having a system comprises: a sensing unit (2) which is placed in the vehicle or mobility and obtains the FOD information of the road in real time while driving on the road; an operation unit (4) which collects raw data received from the sensing unit (2), classifies the data, extracts an object, vectorizes and processes the extracted object data, removes noise through preprocessing, detects an object by an FOD detection modeled algorithm, classifies the risk of an abnormal object into 15 levels, and infers the value of FOD information for the risk level; an output unit (11) for displaying the recognized abnormal object on a navigation system; a communication unit (10) which transmits the processed FOD information to a central control center and propagates the information to surrounding vehicles (1); and a control unit (8) which controls the components to perform autonomous artificial intelligence abnormal object recognition. The present invention prevents a safety accident and enables efficient autonomous driving.

Description

The settlement method and system of expressing FOD on the road for artificial intelligence autonomous driving and the information usage charge by inferring the information value according to the risk of each abnormal object}

The present invention relates to an artificial intelligence-based autonomous vehicle, and more particularly, collects road image data for a real-time location on a road based on GIS and GPS, pre-processes it, and analyzes it in real time. and Debris (abnormal object on the road) information is classified according to the level of risk, and it is displayed on the driver's navigation system, and FOD information is shared with surrounding vehicles and service subscribers following the rear from the collection point where FOD was discovered, thereby preventing accidents. , it relates to an autonomous driving service technology that can generate revenue by charging a fee for information provision and distributing it in a certain ratio between the information provider and the user.

In general, an autonomous car refers to a car that moves by itself without a driver driving the vehicle.

The autonomous vehicle is equipped with various sensors and cameras, for example, RADAR (Radio detection and ranging), LIDAR (Light detection and ranging), 3D shape recognition sensor, camera, laser sensor, multi-axis motion sensor, etc. are installed. do.

Radar is for vehicle and road facility detection, and 2.4GHz short-range radar and 77-78GHz mid-range radar are mainly used. It is mainly used for front and rear collision warning and collision avoidance systems.

Lidar irradiates a laser pulse to the ground surface and obstacles and analyzes the reflected light energy to recognize 3D information around it.

The 3D shape recognition sensor is an image sensor camera with LED and recognition pixels, and is used for omnidirectional collision prevention, lane departure prevention, and parking assistance.

The camera converts the light coming through the lens into a digital signal, and it is applied to around view and rear camera, as well as AEB (Automatic Emergency Braking Brake) and LKA (Lane Keeping Assist System).

Radar is a sensor that uses the principle that radio waves are transmitted from a transmitter, reflected off an obstacle, and then received by a receiver, and is applied to determine distance, speed, and angle.

A multi-axis motion sensor is a sensor that detects the translation or rotation of a vehicle, and is used for posture control of vehicles such as ABS and ESP.

FOD: Foreign Object and Debris refers to various objects that may interfere with driving and cause safety accidents while a vehicle is traveling on a road.

For example, the ideal object includes a human, an elk, a wild boar, a dog, and a cat as mammals, and the object is a rockfall on the road such as gravel and rocks, various vehicle fragments such as tire fragments, bumper fragments, lamp fragments, bolts, etc. It refers to various vehicle cargo parts such as , nuts and pieces, household waste such as cans, plastic vinyl, PET bottles, and glass bottles, and natural phenomena such as black ice, water puddles, and potholes.

And, the road information obtained by these various sensors can be analyzed by the control unit to determine whether there are dangers and obstacles, and also, such big data can be transmitted to a central control center through a communication network such as 5G.

However, such a conventional autonomous vehicle has problems in that it takes a long time and is inaccurate to detect and recognize a front obstacle.

Korean Patent Laid-Open Patent No. 10-2016-0034719 (Name: LiDAR system)

Accordingly, the present invention has been devised to solve such a problem, and an object of the present invention is to obtain foreign object and debris (FOD) data on a road from a road image for a real-time location on a road based on GIS and GPS. It collects, pre-processes, analyzes it in real time, and shares it with nearby vehicles and service subscribers who want to receive information on real-time dangerous situations on the road such as FOD, thereby preventing safety accidents and providing technology for efficient autonomous driving. will do

Another object of the present invention is to provide an autonomous driving service technology capable of generating profits by charging a usage fee according to information provision and distributing it in a certain ratio between the information provider and the user.

The present invention has been proposed to achieve the above object, an embodiment of the present invention,

a sensing unit 2 disposed in a vehicle or mobility to obtain information on FOD data on a road in real time;

Collecting raw data received from the sensing unit 2, classifying and extracting objects, processing the extracted object data as a vector, removing noise through preprocessing, and applying the FOD detection modeled algorithm an operation unit (4) for detecting an object by means of a method, classifying the risk level of the abnormal object into 15 stages, and inferring the value of the FOD information for the risk stage;

an output unit 11 for displaying the recognized abnormal object in the navigation;

a communication unit 10 that transmits the processed FOD information to the central control station and propagates to the surrounding vehicles 1;

Provided is an artificial intelligence-based autonomous driving vehicle (1) including a control unit (8) that controls the above components to perform autonomous driving artificial intelligence abnormal object recognition.

As described above, the AI-based autonomous vehicle according to the present invention collects road image data for a real-time location on the road based on GIS and GPS, pre-processes it, and analyzes it in real-time, Sharing with vehicle and service subscribers has the effect of preventing safety accidents and enabling efficient autonomous driving.

In addition, by differentially classifying risk information by risk level and providing it to surrounding vehicles, there is an effect of generating profits by charging a usage fee according to the information provision and distributing the information at a certain ratio between the information provider and the user.

1 is a perspective view schematically showing the structure of an artificial intelligence-based autonomous driving vehicle according to an embodiment of the present invention.
FIG. 2 is a diagram schematically illustrating a structure in which the autonomous driving vehicle shown in FIG. 1 is linked with another autonomous driving vehicle through a network.
3 is a block diagram schematically showing each configuration of the autonomous vehicle shown in FIG. 1 .
FIG. 4 is a block diagram schematically showing the structure of the operation unit shown in FIG. 3 .
FIG. 5 is a diagram schematically showing a structure for analyzing the degree of risk by the abnormal object determining unit shown in FIG. 4 .
6 is a diagram schematically showing a connection structure of the preprocessing unit and the abnormal object determining unit shown in FIG. 4 .
7 is a block diagram schematically illustrating a process of receiving and processing a signal by the sensing unit shown in FIG. 3 .
FIG. 8 is a block diagram schematically showing the structure of the risk information charging and settlement unit shown in FIG. 3 .
9 is a flowchart illustrating a process in which an autonomous vehicle travels by sensing a road periphery by a sensing unit according to an embodiment of the present invention.
10 is a flowchart illustrating a process of driving an autonomous vehicle by acquiring video data around a drawing by a camera according to an embodiment of the present invention.

Hereinafter, an artificial intelligence-based autonomous driving vehicle according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 to 8, the artificial intelligence-based autonomous driving vehicle 1 proposed by the present invention includes: a sensing unit 2 disposed in a vehicle or mobility to obtain traffic information about a road;

Collecting, classifying and correcting raw data received from the sensing unit 2, extracting an object, vectorizing and processing the extracted object data, removing noise through preprocessing, and modeling algorithm an operation unit (4) for detecting an object by learning data and determining an abnormal object;

an output unit 11 for displaying the recognized abnormal object in the navigation;

a communication unit 10 that transmits the processed traffic information to the central control center and propagates it to the surrounding vehicles 1; a control unit (8) for controlling the components to perform autonomous driving artificial intelligence abnormal object recognition; and a storage unit 12 for storing data.

In this artificial intelligence-based autonomous vehicle (1),

The sensing unit 2 includes real-time road images, vehicles, foreign objects and debris (FODs) on the road and objects around the road, real-time road location data based on GIS and GPS, speed, inter-vehicle distance, images, sound signals, etc. to detect

The autonomous vehicle is provided with various sensors and cameras 7 , for example, RADAR (Radio detection and ranging; 5), LIDAR (Light detection and ranging; 3), 3D shape recognition sensor, and camera 7 , laser sensor, multi-axis motion sensor, etc. are installed.

And, the data obtained by the sensor and camera 7 is pre-processed by the receiving module 30 and then transmitted to the calculating unit 4 .

The reception module 30 includes a reception signal generation module 32 for receiving signals transmitted from LIDAR, RADAR, and a camera, a classification unit 34 for classifying the reception signal, and a generation unit 36 for correcting and generating the signal. ) is included.

Accordingly, the data processed by the receiving module 30 is transmitted to and processed by the operation unit 4 .

The calculation unit 4 includes, as shown in FIG. 4 , a data collection unit 20 for collecting raw data transmitted from the sensing unit 2 ;

Classifying and correcting the signal data received in real time through the data collection unit 20, analyzing the input image to extract a valid image, extracting object data and location information at the same time, and vectorizing the extracted one or more object data an extracting unit 22 for image processing and reception data;

a preprocessor 24 for removing reflected light, angle of view, dust, and noise from the abnormal object data processed by the extraction unit 22;

a detection unit 26 for detecting an object by determining a degree of similarity between the 3D image data processed in the preprocessor 24 and the object data learned by an algorithm modeled from the extracted image data;

and an abnormal object determining unit 26 that registers the detected one or more object data in the database or compares and analyzes the detected abnormal object with the previously registered abnormal object to determine whether it is an abnormal object.

In this calculation unit 4,

The data collection unit 20 includes road data obtained by various sensors and cameras 7, for example, time road images, vehicles, foreign objects and debris (FODs) on the road, and data on objects around the road; It collects raw data for real-time location data on the road based on GIS and GPS.

In addition, the sensor and image data collected by the data collection unit 20 are transmitted to the extraction unit 22 .

The extraction unit 22 classifies and corrects sensor data and image data received in real time through the data collection unit 20 .

Then, the input image is analyzed to extract a valid image, and object data and location information are simultaneously extracted. In addition, one or more extracted object data are vectorized and function-processed.

Then, the extracted object data is transmitted to the pre-processing unit 24 and pre-processed, so that reflected light, angle of view, dust, and noise are removed.

That is, in the raw data obtained by the extraction unit 22, the preprocessing unit 24 requires a preprocessing process due to a structure inappropriate for analysis, missing items, and missing values.

Therefore, through this preprocessing process, noise removal, duplicate value removal, missing value correction, data linkage/integration, data structure change (dimensional change), data vectorization, outlier, and feature engineering are performed. proceed

And, the image and visualization data of the abnormal object recognized by the algorithm modeled by the learned data are the abnormal objects and abnormalities on the road in a state in which the elements obstructing the view are removed even in bad weather conditions such as rain, fog, and snow. You can share object visualization data.

The sensor and image data (hereinafter, traffic data) processed by the pre-processing unit 24 are transmitted to the detection unit 26 and the object is detected by determining the similarity by the modeled algorithm.

The detection unit 26 performs tasks for data prediction, classification, and regression in order to detect an object. That is, object data can be obtained by dividing the preprocessed data by viewpoint.

For example, the vehicle model can be identified by analyzing the data on the outline of the vehicle 1 before and after traveling on the road by an algorithm to determine the degree of similarity.

In this way, by analyzing the traffic data and determining the degree of similarity, objects such as the shape of the road surface and obstacles around the road are identified.

On the other hand, the detected object data is compared and analyzed with the previously registered abnormal object by the abnormal object determination unit 26 to determine whether the object is an abnormal object.

The abnormal object determination unit 26 recognizes the surrounding vehicle 1, obstacles, and lanes (hereinafter, traffic data) by the built-in artificial intelligence and determines whether there is a danger.

This abnormal object determination unit 26 is an AI core such as machine learning and deep learning, an AI framework such as Tensorflow, SSD, YOLO, an Android or Linux-based OS, and a processor such as CPU, GPU, and TPU. include

In addition, the abnormal object determination unit 26 supports at least 8 channels or more, and as an interface, USB, HDMI, Ethernet 2, 4G/5G dual-band, Bluetooth, and the like are provided.

The abnormal object determination unit 26 recognizes traffic data such as surrounding vehicles 1, obstacles, and lanes by analyzing the processed data. Such an abnormal object determination unit 26 may be processed by a central processing unit (CPU) 27 equipped with artificial intelligence.

As shown in FIG. 6 , the central processing unit 27 includes an abnormal object analysis engine, a GPU, an ISP (Image Signal Processor), an ALSA (Advanced Linux Sound Architecture), an RTSP (Real Time Streaming Protocol), and a communication module. signals are transmitted and received by

The abnormal object determination unit 26 may analyze real-time road information using various algorithms, for example, it is a method of analyzing by a deep learning method.

In other words, deep learning is a machine learning method built on the basis of a deep neural network (CNN) so that a computer can learn by itself like a human using multiple data.

When a deep neural network is used, classification and clustering of traffic data are possible, and similarity judgment can be performed by placing various layers on the data for which classification or clustering is desired.

That is, the degree of similarity can be determined by extracting shape features of traffic data with an artificial neural network and classifying or clustering them by shape using the features as input values for other machine learning algorithms.

The artificial neural network includes a deep neural network, and the deep neural network refers to a neural network composed of several layers among neural network algorithms.

That is, a deep neural network is composed of multiple layers, and each layer is composed of several nodes, and an operation to actually classify a shape occurs at each node. designed to do

When a node receives a stimulus of a certain size or more, it responds, and the magnitude of the response is approximately proportional to the product of the input value and the node's coefficient (or weights).

In general, a node receives multiple inputs and has as many coefficients as the number of inputs. Therefore, different weights can be given to different inputs by adjusting this coefficient.

Finally, all the multiplied values are added and the sum is input to the activation function. The result of the activation function corresponds to the output of the node, and this output is ultimately used for classification or regression analysis.

Each layer consists of several nodes, and whether each node is activated/deactivated is determined according to the input value. At this time, the input data becomes the input of the first layer, and after that, the output of each layer becomes the input of the next layer again.

All coefficients change little by little during the shape learning process of traffic data, and as a result, reflect which input each node considers important. And training of the neural network is the process of updating this coefficient.

When learning the shape of traffic data, in this deep neural network, features of different layers are learned for each layer.

That is, simple and specific features are learned for low-level features (eg, the shape constituting the shape of the vehicle 1), and more complex and abstract features are learned from high-level features (eg, inter-vehicle distance, surrounding guardrails)

Through this abstraction learning process, the deep neural network understands high-dimensional data, and hundreds of millions of coefficients are involved in this process. (A non-linear function is used in this process.)

In addition, deep neural networks can use the data to identify the latent structures of the data. That is, potential structures such as the shape of the surrounding vehicle 1 , the inter-vehicle distance, guard rails, lanes, paint lanes, and pedestrians can be grasped. Through this, similarity between data can be effectively identified even when data is not labeled, and as a result, deep neural networks are effective for clustering traffic data.

For example, by using a neural network, a large amount of traffic data can be input, and similar objects can be grouped and classified. That is, objects such as a vehicle, a traffic signboard, and a pedestrian may be classified into similar objects.

The difference between this deep neural network and general machine learning is that feature extraction is performed automatically.

In the past, in order to extract effective features, experts in the relevant field had to devise and apply formulas or methods for extracting features directly for a long time. This method has a problem in that it takes a lot of time to develop, evaluate, and supplement.

A deep neural network is an algorithm designed to make a computer take over this process, and it trains it to perform much faster and more effectively than humans.

Even when learning unlabeled data, neural networks can automatically extract features of traffic data. There are several methods for this automatic extraction, and in general, this process learns to make the output equal to the input when passed through the neural network.

No matter what kind of label is (use the input as it is/use a separate label), the neural network finds the correlation between the input and the output. In some cases, after training the neural network to some extent with labeled data, it is possible to continue learning by adding unlabeled data. By using this method, the performance of the neural network can be maximized.

The last layer of a deep neural network is the output layer. In most cases, the activation function of the output layer is logistic or softmax, and the probability of a specific label can be finally obtained from the output layer. For example, when traffic data is input, whether the vehicle is small, medium, large, van, or bus, the lane is straight or curved, the presence or absence of pedestrians, etc. can be obtained with respective probabilities.

First, all coefficients of the neural net are initialized before training starts. Then, the learning proceeds by repeatedly inputting traffic data. If the learning proceeds smoothly, the coefficients will be updated to appropriate values, and various classifications and predictions are possible with this artificial neural network.

In the learning process, the updating process of these coefficients occurs repeatedly.

The principle of coefficient update is to first estimate the coefficient, measure the error that occurs when the coefficient is used, and then slightly update the coefficient based on the error.

At this time, the multiple coefficients of the neural network are collectively called a model, and the model may be in an initialized state or in a state in which learning is completed.

The initialized model does not perform meaningful work, but as the learning progresses, the model outputs results similar to the actual value rather than a random value.

This is because the artificial neural network does not know anything before data is input, and the reason for initializing coefficients to random values is also the same. And as the data is read, the coefficients are updated little by little in the correct direction.

Through this update process, the artificial neural network can group similar traffic data by classifying the input traffic data.

The traffic data clustered in this way is registered in the database 12 .

The abnormal object determining unit 26 can recognize the objects on the road in this way, and determines whether there is a danger between the recognized objects and the present vehicle 1 .

That is, the abnormal object determination unit 26 determines whether there is a danger by the risk determination module 29, and the distance, speed, travel direction, lane maintenance, and obstacle position between the present vehicle 1 and the surrounding vehicles 1 By recognizing it, it is possible to determine whether there is a risk or not.

For example, the risk determination module 29 compares the traveling speed of the present vehicle 1 with the speed and distance of the front vehicle 1 or the rear vehicle 1, and also compares the traveling direction.

As a result of comparison, when the speed difference between the front vehicle 1 and the present vehicle 1 is less than a certain range, the distance is also within a certain distance, and the traveling direction is the same, it is determined that both vehicles 1 have a high probability of collision.

In this case, the risk determination module 29 operates a braking device by transmitting a signal to the control unit 8 of the vehicle 1 to decelerate the vehicle 1 .

In addition, when the obstacle on the road ahead is recognized, the risk determination module 29 compares the traveling direction, speed, and distance of the vehicle 1 . As a result of comparison, if the traveling direction of the vehicle 1 is located in the same direction as the obstacle and the distance from the obstacle is within a certain range, it is determined that it is dangerous and transmits a signal to the control unit 8 to operate the braking device to decelerate.

At this time, the abnormal object determination unit 26 does not specifically specify objects such as the vehicle 1 and obstacles on the road, and does not form a DB, and the location information based on GIS and GPS and the position and characteristics (shape) of the abnormal object , size, type) are recognized and analyzed.

In addition, the level of risk is divided into multiple stages, for example, 15 stages, through situation analysis such as the size and location of the abnormal object, and the value of the information is calculated for each stage by classifying the abnormal object.

At this time, it is calculated by inferring the value standard for each risk information for the abnormal object detected in real time.

Various methods are available for inferring the value of risk information. For example, it is a method of inferring the value of risk by storing data about a dangerous situation in a DB and comparing it with this data. At this time, value inference is performed by artificial intelligence such as deep learning, and may be performed, for example, by a decision tree.

In addition, the first discoverer of the FOD information resets the sensor unit for a predetermined period of time, for example, every 15 minutes to continuously update in real time, and repeats until the FOD is removed.

Meanwhile, the FOD object data recognized by the abnormal object determining unit 26 may be displayed through the output unit 11 such as a navigation system. In this case, the output unit 11 includes not only a navigation device, but also a laptop computer, a tabbook, a PC, a smart phone, and the like.

Accordingly, the user can grasp the dangerous situation of the road in real time through the traffic object related to road driving, such as the road, the vehicle 1, and the FOD displayed on the navigation system.

Then, the communication unit 10 transmits the real-time dangerous situation of the processed road to the central control office and propagates it to the surrounding vehicles 1 .

That is, the central control office transmits real-time road condition information through the network to a nearby vehicle (1) that agrees to receive real-time risk information on the road by agreeing to receive real-time risk information around the vehicle 1 or by transmitting it to a vehicle registered as a member. ) can grasp the real-time dangerous situation of the road in the relevant area.

At this time, the first detection vehicle of the FOD data is one vehicle determined by the detection time of the data and the priority order of arrival of the FOD information to the central center, and within a certain distance from the rear based on the position detected by the detection vehicle, for example, within 2Km Transmits danger information to nearby vehicles.

In addition, when the level of risk is higher than a predetermined level, for example, when it is determined that the level is higher than 10, the location information of FOD and FOD is uploaded or downloaded in real time, so that customers or members who want to receive information from vehicles following around It propagates to the vehicle (1).

On the other hand, in the present invention, when the risk information is disseminated to other surrounding drivers by providing the risk information billing and settlement unit 40, the information value according to the degree of risk can be converted into a cost and claimed.

That is, the risk information billing and settlement unit 40 is connected to the control unit 8, and the risk information billing and settlement unit 40 is a risk determined by the abnormal object determining unit 26, as shown in FIG. a risk rating module 42 for grading; a vehicle recognition module 44 for receiving the location and identification information of surrounding vehicles from the communication unit 10; an information providing module 46 for providing danger information to the recognized surrounding vehicles; a charging module 48 for charging a usage fee according to the risk information provided to the owner of the surrounding vehicle; and a settlement module 50 for distributing the usage fee charged by the billing module 48 as a profit ratio.

In this risk information billing and settlement unit,

The rating module 42 classifies the road conditions determined by the abnormal object determination unit 26 into a plurality of classes according to the degree of risk, for example, into 15 classes.

The vehicle recognition module 44 acquires recognition information of vehicles located in the vicinity from the communication unit 10 . In this case, the recognition information includes a car number or a discrimination safety code. Of course, such identification information is a case in which the borrowers have agreed in advance to provide the information.

And, the information providing module 46 provides risk information to the recognized surrounding vehicles. In this case, the risk information includes road delay and congestion, accident location, accident lane, accident vehicle type, congestion situation at the accident point, whether there are obstacles, and the like. This risk information is updated in real time.

Then, the charging module 48 charges a usage fee according to the risk information to the surrounding vehicles to which the risk information is provided.

That is, when the borrowers who have signed up for membership in advance receive risk information, a certain fee is charged according to the information provided. In this case, each borrower may be provided with only risk information suitable for the borrower by setting the information desired to be provided in advance.

In addition, the risk information may differentially set usage fees according to the information.

For example, the usage fee is charged by differentiating general traffic information, specific accident information such as an accident location, and risk information such as whether obstacles appear.

In addition, the usage fee may be paid by a conventional payment method, and may be paid by online payment, account remittance, membership fee payment, or the like.

In this way, a certain usage fee can be charged to surrounding vehicles by the risk information billing and settlement unit 40 , and profits can be realized by settlement by the settlement module 50 .

At this time, the distribution ratio of the profit can be set in various ways, for example, the service operator can distribute the distribution at a rate of 10 to 40% and the driver providing risk information at a rate of 60 to 90%.

The calculation unit 4, the control unit 8, the risk rating module 42, the vehicle recognition module 44, the information providing module 46, the charging module 48, and the settlement module 50 are It refers to a microprocessor that receives signals and sequentially performs computer control, operation, and memory processing.

Hereinafter, the process of driving the autonomous driving vehicle based on artificial intelligence of the present application by radar, lidar and camera will be described in more detail.

As shown in FIG. 9 , when the autonomous vehicle 1 is driving, the current location data by the sensing unit 2 and the surrounding road environment are scanned by the lidar 3 and radar 5 first.

Then, the distance between vehicles, the distance to the object, and the stereoscopic image of the object on the road are measured by the sensing unit 2 .

It is determined whether the data measured in this way belongs to the data category modeled by the detector 26 .

If it belongs to the modeled category, 3D modeling is performed, and abnormal objects are identified by the modeled artificial intelligence.

On the other hand, if it does not belong to the modeled category, it is determined whether or not deep learning is performed.

If the judgment result is judged learning, data is transmitted to the server to perform 3D modeling and deep learning. And classify the object and make DB.

If deep learning is not carried out, class classification of objects and DB are performed directly.

As a result of determining the abnormal object in the above step, it is determined whether the abnormal object is dangerous to driving in 15 steps. At this time, it is calculated by inferring the value standard for the information of the abnormal object detected in real time.

As a result of the judgment, if it is determined that it is dangerous, the risk factor is removed by driving at a low speed or performing a detour. Alternatively, the risk factor is removed by stopping the vehicle 1 when driving at low speed or not detouring.

For real-time FOD information indicating the risk level of 10 or higher risk stages as a result of judgment, upload or download FOD and FOD location information in real time and disseminate it to vehicles (1) of customers or members who want to receive information from among vehicles following do.

Meanwhile, as shown in FIG. 10 , the autonomous vehicle 1 may drive according to image information acquired by the camera.

That is, real-time road image information is obtained by scanning the road environment by the camera, and current location data is obtained.

Then, it is determined whether the image information belongs to the data category modeled by the detector 26 .

If it belongs to the modeled category, pre-processing of image information is performed, and dust such as reflected light and angle of view is removed and corrected.

Then, the image information from which the dust has been removed is identified by the modeled artificial intelligence as an abnormal object.

On the other hand, if it does not belong to the modeled category, it is determined whether or not deep learning is performed.

If the judgment result is judged learning, the data is transmitted to the server and deep learning is carried out. And classify the object and make DB.

If deep learning is not carried out, class classification of objects and DB are performed directly.

As a result of determining the abnormal object in the above step, it is determined whether the abnormal object is dangerous for driving by dividing it into a plurality of steps, for example, 15 steps. At this time, for each abnormal object detected in real time, the value standard for risk information is inferred and calculated.

As a result of the judgment, if it is determined that it is dangerous, the risk factor is removed by driving at a low speed or performing a detour. Alternatively, the risk factor is removed by stopping the vehicle 1 when driving at low speed or not detouring.

For real-time FOD information indicating the risk level of 10 or higher risk stages as a result of judgment, upload or download FOD and FOD location information in real time and disseminate it to vehicles (1) of customers or members who want to receive information from among vehicles following do.

Meanwhile, according to the present invention, when the risk information on the road is transmitted to other drivers around the road by adding the risk information charging step, the information value according to the degree of risk is converted into a cost and a user fee is charged to realize profits.

That is, the risk information billing and settlement unit 40 is connected to the control unit 8 , and the risk information billing and settlement unit 40 receives risk information from the abnormal object determining unit 26 .

And, the road condition determined by the abnormal object determination unit 26 is divided into a plurality of grades according to the degree of risk by the grade module 42 . For example, it is divided into 15 grades.

Then, the vehicle recognition module 44 acquires recognition information of vehicles located in the vicinity from the communication unit 10 . In this case, the recognition information includes a car number or a discrimination safety code. Of course, such identification information is a case in which the borrowers have agreed in advance to provide the information.

Then, the information providing module 46 provides risk information to the recognized surrounding vehicles. In this case, the risk information includes road delay and congestion, accident location, accident lane, accident vehicle type, congestion situation at the accident point, whether there are obstacles, and the like. This risk information is updated in real time.

Then, the charging module 48 charges a usage fee according to the risk information to the surrounding vehicles to which the risk information has been provided.

That is, when the borrowers who have signed up for membership in advance receive risk information, a certain fee is charged according to the information provided. In this case, each borrower may be provided with only risk information suitable for the borrower by setting the information desired to be provided in advance.

In addition, the risk information may differentially set usage fees according to the information.

For example, a fee is charged by differentiating general traffic information, specific accident information such as an accident location, and risk information such as whether obstacles appear.

In addition, the usage fee may be paid by a conventional payment method, and may be paid by online payment, account remittance, membership fee payment, or the like.

In this way, the risk information billing and settlement unit 40 provides information to surrounding vehicles and charges a certain usage fee, and the settlement module 50 can realize profits.

In this case, the distribution ratio of the profit can be set in various ways, for example, the service operator can decompose it at a rate of 10 to 40% and the driver providing risk information at a rate of 60 to 90%.

Claims (9)

a sensing unit (2) disposed in a vehicle or mobility to obtain information on abnormal object data on a road in real time;
Collects raw data received from the sensing unit 2, classifies and extracts objects, processes the extracted object data as vectors, removes noise through preprocessing, and detects abnormal objects Modeled algorithm an operation unit (4) for detecting an object, classifying the risk of the abnormal object into 15 stages, and inferring the value of the abnormal object information for the dangerous stage;
an output unit 11 for displaying the recognized abnormal object in the navigation;
a communication unit 10 that transmits the processed abnormal object information to the central control center and propagates it to the surrounding vehicles 1;
It includes the sensing unit (2), the calculating unit (4), the output unit (11), and the control unit (8) for controlling the communication unit (10) to proceed with autonomous driving artificial intelligence abnormal object recognition,
The calculating unit 4 includes: a data collecting unit 20 for collecting raw data transmitted from the sensing unit 2; Classifying and correcting the signal data received in real time through the data collection unit 20, analyzing the input image to extract a valid image, extracting object data and location information at the same time, and vectorizing the extracted one or more object data an extracting unit 22 for image processing and reception data; a pre-processing unit 24 for removing noise from the abnormal object data processed by the extraction unit 22; a detection unit 26 for detecting an object by determining a degree of similarity between the 3D image data processed in the preprocessor 24 and the object data learned by an algorithm modeled from the extracted image data; and an abnormal object determination unit 26 for registering one or more detected object data in the database or comparing and analyzing it with previously registered abnormal objects to determine whether an abnormal object is present,
A risk information billing and settlement unit 40 is connected to the control unit 8, and the risk information billing and settlement unit 40 includes a rating module 42 for grading the risk determined by the abnormal object determining unit 26 and ; a vehicle recognition module 44 for receiving the location and identification information of surrounding vehicles from the communication unit 10; an information providing module 46 that provides danger information to the recognized surrounding vehicles; An artificial intelligence-based autonomous driving vehicle (1) including a charging module (48) that charges a usage fee according to the risk information provided to the owner of the surrounding vehicle. The method of claim 1,
The sensing unit 2 is based on artificial intelligence including one of a radio detection and ranging (RADAR) 5, a light detection and ranging (LIDAR) 3, a 3D shape recognition sensor, a camera 7, a laser sensor, and a multi-axis motion sensor. of autonomous vehicles (1). delete The method of claim 1,
The abnormal object determination unit 26 is processed by a central processing unit (CPU; 27) equipped with artificial intelligence,
Central processing unit 27, an abnormal object analysis engine on the road, GPU, ISP (Image Signal Processor), ALSA (Advanced Linux Sound Architecture), ALVA, RTSP (Real Time Streaming Protocol), communication module, RAM, LAN Artificial intelligence-based autonomous vehicle (1) comprising a. The method of claim 1,
The extraction unit 22 compares and corrects the coordinate values and position values scanned with lidar (3) and radar (5) with the real image data inputted by the camera (7), an artificial intelligence-based autonomous vehicle (1). The method of claim 1,
The abnormal object determination unit 26 recognizes only roads and cars as objects, and the other objects are recognized as abnormal objects. The method of claim 1,
The data obtained by the sensing unit 2 is pre-processed by the receiving module 30 and then transmitted to the calculating unit 4,
The reception module 30 includes a reception signal generation module 32 for receiving signals transmitted from LIDAR, RADAR, and a camera, a classification unit 34 for classifying the reception signal, and a generation unit 36 for correcting and generating the signal Artificial intelligence-based autonomous vehicle (1) comprising a. delete The autonomous driving vehicle (1) based on artificial intelligence (1) according to claim 1, wherein the charging module (48) sets the usage fee differentially according to the information of the risk information to charge the usage fee.

Images