Method for control of autonomous vehicle and device for executing the method

Abstract

A method for controlling an autonomous vehicle and an apparatus for performing the same are disclosed. A method for controlling an autonomous vehicle and an apparatus for performing the same according to an embodiment of the present invention comprises: an image sensor module for acquiring original image information capturing the front of an autonomous vehicle; a communication module for acquiring external image information from surrounding autonomous vehicles; a feature extraction module for extracting an original image frame from the original image information and extracting feature vector information for a plurality of objects within the original image frame, respectively, and extracting an external image frame from the external image information and extracting feature vector information for a plurality of objects within the external image frame, respectively; a fusion module for fusing the original image frame and the external image frame based on the feature vector information for a plurality of objects within the original image frame and the feature vector information for a plurality of objects within the external image frame; an identification module for identifying a hidden object based on the fused image frame; a judgment module for determining a risk level based on the risk of collision between the identified hidden object and the autonomous vehicle; and a driving control module for controlling the speed of the autonomous vehicle according to the determined risk level.

Description

Technology Field

[0001] Embodiments of the present invention relate to technology for controlling autonomous vehicles. Background Technology

[0003] An autonomous vehicle is a vehicle capable of detecting its surroundings and movements with little to no human intervention or input. For such autonomous vehicles to be realized, various sensors capable of sensing surrounding conditions must be combined, such as radar, computer vision, GPS, odometry, and inertial measurement units. These autonomous vehicles observe surrounding road conditions using various sensors, including radar and cameras.

[0004] However, if specific vehicles or obstacles are present within the sensor area detected by an autonomous vehicle, blind spots may occur where such vehicles or obstacles are not detected. Furthermore, since conventional autonomous vehicles typically display only detected vehicles or objects to the driver and do not show undetected blind spots, there is a problem in that it is difficult for the driver to pay attention to these blind spots. Additionally, there is a problem in determining and sensing the forward position to be detected, as the vehicle is unable to detect obscured objects and obstacles located beyond the field of view. Prior art literature

[0006] Republic of Korea Published Patent Application No. 10-2011-0097091 (August 31, 2011) The problem to be solved

[0007] Embodiments of the present invention are intended to improve the stability of an autonomous vehicle by detecting hidden objects beyond the field of view that cannot be detected. means of solving the problem

[0009] According to an exemplary embodiment of the present invention, a control device for an autonomous vehicle is provided, comprising: an image sensor module for acquiring original image information captured in front of an autonomous vehicle; a communication module for acquiring external image information from surrounding autonomous vehicles; a feature extraction module for extracting an original image frame from the original image information and extracting feature vector information for a plurality of objects within the original image frame, and for extracting an external image frame from the external image information and extracting feature vector information for a plurality of objects within the external image frame; a fusion module for fusing the original image frame and the external image frame based on the feature vector information for a plurality of objects within the original image frame and the feature vector information for a plurality of objects within the external image frame; an identification module for identifying a hidden object based on the fused image frame; a judgment module for determining a risk level based on the risk of collision between the identified hidden object and the autonomous vehicle; and a driving control module for controlling the speed of the autonomous vehicle according to the determined risk level.

[0010] The above feature extraction module can extract feature vector information for multiple objects within the original image frame and multiple objects within the external image frame through SIFT (Scale-Invariant Feature Transform).

[0011] The fusion module can determine the similarity between feature vector information for multiple objects of the original image frame and feature vector information for multiple objects of the external image frame, and perform matching between objects determined to have high similarity to fuse the original image frame and the external image frame.

[0012] The above feature vector information includes the position, orientation, size, and gradient vector of an object, and the fusion module calculates the Euclidean distance between the gradient vectors of multiple objects in the original image frame and the gradient vectors of multiple objects in the external image frame, and can determine that objects with a distance less than or equal to a preset distance have high similarity.

[0013] The identification module can compare the fused image frame with the original image frame to detect an object present in one image frame, and identify the detected object to set it as a hidden object.

[0014] The above identification module can identify the detected object by an object recognition method based on Faster R-CNN (Regions with Convolutional Neuron Networks).

[0015] The feature extraction module described above acquires the original image information and the external image information based on previously stored road information and original vehicle information acquired from the autonomous vehicle, when the distance between the autonomous vehicle and the crosswalk is within a preset distance, and can acquire the external image information from another autonomous vehicle within a preset distance in the direction of travel of the autonomous vehicle.

[0016] The above judgment module calculates the estimated time of arrival to the crosswalk based on original vehicle information obtained from the autonomous vehicle, calculates the estimated time of crossing to the sidewalk based on the movement information of the hidden object, determines the risk level by comparing the estimated time of arrival and the estimated time of crossing, and the movement information of the hidden object can be generated by extracting the position, speed, and direction of travel of the hidden object from consecutive frames of the fused image frames. Effects of the invention

[0018] According to embodiments of the present invention, the stability of an autonomous vehicle can be improved by sharing images of surrounding autonomous vehicles to identify objects located in blind spots. Brief explanation of the drawing

[0020] FIG. 1 is a block diagram illustrating the configuration of a control device for an autonomous driving vehicle according to an embodiment of the present invention. FIG. 2 is a drawing to explain the process of a blind spot being formed at a level crossing. FIG. 3 is a flowchart illustrating a control method for an autonomous driving vehicle according to an embodiment of the present invention. FIG. 4 is a block diagram illustrating a computing environment including a computing device suitable for use in exemplary embodiments. Specific details for implementing the invention

[0021] Hereinafter, specific embodiments of the present invention will be described with reference to the drawings. The following detailed description is provided to facilitate a comprehensive understanding of the methods, apparatuses, and / or systems described herein. However, this is merely illustrative and the present invention is not limited thereto.

[0022] In describing the embodiments of the present invention, detailed descriptions of known technologies related to the present invention are omitted if it is determined that such detailed descriptions may unnecessarily obscure the essence of the present invention. Furthermore, the terms described below are defined in consideration of their functions within the present invention, and these may vary depending on the intentions or practices of the user or operator. Therefore, such definitions should be based on the content throughout this specification. Terms used in the detailed description are intended merely to describe the embodiments of the present invention and should not be limiting in any way. Unless explicitly stated otherwise, expressions in the singular form include the meaning of the plural form. In this description, expressions such as "include" or "comprise" are intended to refer to certain characteristics, numbers, steps, actions, elements, parts thereof, or combinations thereof, and should not be interpreted to exclude the existence or possibility of one or more other characteristics, numbers, steps, actions, elements, parts thereof, or combinations thereof other than those described.

[0023] In the following description, terms such as "transmission," "communication," "transmission," "reception," and other terms of similar meaning regarding signals or information include not only the direct transfer of signals or information from one component to another but also the transfer through other components. In particular, "transmission" or "transmitting" a signal or information to one component refers to the final destination of the signal or information and does not mean a direct destination. The same applies to the "reception" of signals or information. Furthermore, in this specification, two or more data or information are "related" means that if one data (or information) is obtained, at least a portion of another data (or information) can be obtained based thereon.

[0024] Meanwhile, embodiments of the present invention may include a program for performing the methods described herein on a computer, and a computer-readable recording medium containing said program. The computer-readable recording medium may include program instructions, local data files, local data structures, etc., either alone or in combination. The medium may be one specifically designed and configured for the present invention, or one that is commonly available in the field of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of said programs may include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0026] FIG. 1 is a block diagram illustrating the configuration of a control device for an autonomous vehicle according to one embodiment of the present invention.

[0027] As illustrated in FIG. 1, a control device (100) of an autonomous vehicle according to one embodiment of the present invention may include an image sensor module (110), a communication module (120), a feature extraction module (130), a fusion module (140), an identification module (150), a judgment module (160), and a driving control module (170).

[0028] Meanwhile, Figure 2 is a diagram illustrating the process of a blind spot being formed at a level crossing.

[0029] Referring to FIG. 2, when the current vehicle (10) is an autonomous vehicle, the current vehicle (10) can detect surrounding traffic conditions using an image sensor module (110).

[0030] Here, the first vehicle (20) can be detected using the image sensor module (110) of the current vehicle (10). However, there may be cases where the sensors mounted on the current vehicle (10) by the first vehicle (20) fail to detect an area beyond the first vehicle (20). That is, as shown in FIG. 2, a blind spot is formed in front of the first vehicle (20) that the current vehicle (10) cannot detect. In addition, at intersections other than crosswalks as shown in FIG. 2, a blind spot may be formed because the area in front of the current vehicle (10), which is located on the same lane as the current vehicle (10), is not detected by the image sensor module (110) of the current vehicle (10).

[0031] As such, a blind spot occurs that is not detected by the image sensor module (110) of the current vehicle (10), making it difficult for the current vehicle (10) to respond quickly to various dangerous situations, such as pedestrians (1) crossing in a hurry.

[0032] To solve this problem, the safety of the autonomous vehicle can be improved by sharing images of surrounding autonomous vehicles (30) to identify objects located in blind spots.

[0033] The image sensor module (110) can acquire image information of the front of the vehicle. For example, the image sensor module (110) may be composed of multiple cameras, and multiple cameras may be installed in a distributed manner to acquire image information of the front of the vehicle. Meanwhile, although the present invention is configured to acquire image information of the front of the vehicle, it is not limited thereto, and image information of the side and rear of the vehicle can also be acquired.

[0034] The communication module (120) can perform V2V (Vehicle To Vehicle) communication by wirelessly transmitting and receiving signals or data with surrounding autonomous vehicles. Since the process of exchanging information between autonomous vehicles in such a vehicle-to-vehicle communication environment is a widely known technology, a detailed explanation thereof will be omitted. The communication module (120) can receive video information and vehicle information from surrounding autonomous vehicles. In addition, the communication module (120) can receive road information from an external server.

[0035] In the present invention, information obtained from an image sensor module (110) is referred to as original image information and original vehicle information, and information received from surrounding vehicles is referred to as external image information and external vehicle information. Here, the image information may be an image taken of the front of the vehicle, and the vehicle information may be the vehicle's speed, vehicle location information (GPS (global positioning system) information), etc.

[0036] The feature extraction module (130) can extract original image frames from original image information obtained from the image sensor module (110) and extract feature vector information for multiple objects in the original image frames based on the extracted original image frames. Additionally, the object feature extraction module (130) can extract external image frames from external image information obtained from the communication module (120) and extract feature vector information for multiple objects in the external image frames based on the extracted external image frames.

[0037] Meanwhile, the feature extraction module (130) can acquire original image information from the image sensor module (110) and acquire external image information from the communication module (120) when the distance of the autonomous vehicle to the crosswalk (or intersection) is within a preset distance based on road information and original vehicle information. At this time, the external image information can be received from another autonomous vehicle within a preset distance in the direction of travel of the autonomous vehicle. Additionally, the original image information and the external image information may be images captured at the same time.

[0038] In an exemplary embodiment, the feature extraction module (130) may use SIFT (Scale-Invariant Feature Transform) as a method for extracting and analyzing feature vectors to fuse original image information and external image information. Here, the SIFT algorithm is a representative algorithm for extracting feature vectors of an image and has the characteristic of being invariant to image rotation, scaling, translation, partial lighting changes, and projective transforms. The SIFT algorithm can extract attributes such as the position, scale, orientation, and gradient vector of features by considering the characteristics of the image. For example, the SIFT algorithm can select objects based on extrema in a scale space generated through the DOG (Difference-of-Gaussian) function and extract feature vector information for the selected objects. Here, the feature vector information may include the position of the object (coordinates of the object within the image frame), the orientation of the object (orientation of the object within the image frame), the size of the object (scale of the image frame), and the gradient vector of the object (image fradient).

[0039] Meanwhile, although the present invention is described as receiving external image information from a surrounding autonomous vehicle to extract second feature vector information, it is not limited thereto, and second feature vector information for a plurality of objects in an external image frame extracted from a surrounding autonomous vehicle may also be received.

[0040] The fusion module (140) can fuse the original image frame and the external image frame based on feature vector information for multiple objects in the original image frame and feature vector information for multiple objects in the external image frame. Specifically, the fusion module (140) can fuse the original image frame and the external image frame by determining the similarity between the feature vector information for multiple objects in the original image frame and the feature vector information for multiple objects in the external image frame, and performing matching between the objects determined to have high similarity.

[0041] In an exemplary embodiment, the fusion module (140) calculates the Euclidean distance between the gradient vectors of multiple objects in the original image frame and the gradient vectors of multiple objects in the external image frame, and for objects that are less than or equal to a preset distance, it can determine that they have similar feature vector information and thus determine that they have high similarity.

[0042] Additionally, the fusion module (140) can fuse the original image frame and the external image frame by performing matching between objects obtained from the original image frame and the external image frame that are determined to have high similarity based on the determined similarity. At this time, the position, orientation, and size of the objects may be considered to perform matching between the objects.

[0043] The identification module (150) can detect an object present in one of the two image frames by comparing the original image frame and the fused image frame. Additionally, the identification module (150) can identify the detected object and set it as a hidden object. At this time, object identification within the image frame can be performed using an object recognition method based on Faster R-CNN (Regions with Convolutional Neuron Networks). Additionally, the objects to be identified may include objects crossing a crosswalk, such as people, bicycles, and strollers. Meanwhile, when the objects to be identified are intersections, the objects to be identified may further include objects with movement such as vehicles, two-wheeled vehicles, motorcycles, buses, trucks, and electric bikes.

[0044] The judgment module (160) can determine a risk level based on the driving information of the autonomous vehicle and the movement information of the hidden object when a hidden object is set in the identification module (150). Specifically, when a hidden object is set in the identification module (150), the judgment module (160) can calculate the estimated time of arrival to the crosswalk based on the driving information of the autonomous vehicle and calculate the estimated time of crossing to the sidewalk based on the movement information of the hidden object. Additionally, the judgment module (160) can determine a risk level according to the risk of collision by comparing the estimated time of arrival and the estimated time of crossing. For example, if the judgment module (160) determines that the risk of collision with the identified object is high as a result of determining the risk of collision, it can assign a high risk level, and if the risk of collision is low, it can assign a low risk level. Here, the driving information of the autonomous vehicle may include the vehicle's location, speed, and direction of travel. At this time, the movement information of the hidden object can be generated by extracting the object's location, speed, and direction of travel from consecutive fused image frames.

[0045] The driving control module (170) can control the speed of the vehicle according to the risk level determined by the judgment module (160). Meanwhile, the driving control module (170) can control the vehicle to maintain its current speed if no hidden object is identified by the identification module (150).

[0046] Additionally, the driving control module (170) can prevent the autonomous vehicle from colliding with an identified object by appropriately adjusting the driving speed based on a preset risk level when controlling the speed of the autonomous vehicle. For example, the driving control module (170) can increase the acceleration / deceleration when a high risk level is determined by the judgment module, and decrease the acceleration / deceleration accordingly when a low risk level is determined.

[0047] FIG. 3 is a flowchart illustrating a control method for an autonomous vehicle according to an embodiment of the present invention. The method illustrated in FIG. 3 may be performed, for example, by the control device of the autonomous vehicle described above. Although the method is described in the illustrated flowchart by dividing it into a plurality of steps, at least some of the steps may be performed in a different order, combined with other steps and performed together, omitted, divided into detailed steps, or performed with one or more steps not illustrated added.

[0048] First, the control device (100) of the autonomous vehicle acquires original image information of the autonomous vehicle and extracts the original image frame (S302).

[0049] Next, the control device (100) of the autonomous vehicle acquires external image information of surrounding autonomous vehicles and extracts an external image frame (S304).

[0050] Next, the control device (100) of the autonomous vehicle extracts feature vector information for objects from the original image frame and the external image frame (S306). Specifically, the control device (100) of the autonomous vehicle can extract feature vector information for multiple objects in the original image frame based on the extracted original image frame. Additionally, the control device (100) of the autonomous vehicle can extract feature vector information for multiple objects in the external image frame based on the extracted external image frame. Meanwhile, the control device (100) of the autonomous vehicle can acquire original image information and acquire external image information when the distance of the autonomous vehicle to a crosswalk (or intersection) is within a preset distance based on road information and original vehicle information. At this time, the external image information can be received from another autonomous vehicle within a preset distance in the direction of travel of the autonomous vehicle.

[0051] In an exemplary embodiment, the control device (100) of an autonomous vehicle may use SIFT (Scale-Invariant Feature Transform) as a method for extracting and analyzing feature vectors to fuse original image information and external image information. Here, the SIFT algorithm is a representative algorithm for extracting feature vectors of an image and has the characteristic of being invariant to image rotation, scaling, translation, partial lighting changes, and projective transforms. The SIFT algorithm can extract attributes such as the position, scale, direction, and gradient vector of features by considering the characteristics of the image. For example, the SIFT algorithm can select objects based on extrema in a scale space generated through the DOG (Difference-of-Gaussian) function and extract feature vector information for the selected objects. Here, the feature vector information may include the position of the object (coordinates of the object within the image frame), the orientation of the object (orientation of the object within the image frame), the size of the object (scale of the image frame), and the gradient vector of the object (image gradient of the image frame).

[0052] Next, the control unit (100) of the autonomous vehicle fuses the original image frame and the external image frame based on the extracted feature vector information (S308). Specifically, the control unit (100) of the autonomous vehicle can fuse the original image frame and the external image frame by determining the similarity between the feature vector information for multiple objects in the original image frame and the feature vector information for multiple objects in the external image frame, and performing matching between the objects determined to have high similarity.

[0053] Next, the control device (100) of the autonomous vehicle identifies a hidden object from the fused image frame (S310). Specifically, the control device (100) of the autonomous vehicle can detect an object present in one of the two image frames by comparing the original image frame and the fused image frame. Additionally, the control device (100) of the autonomous vehicle can identify the detected object and set it as a hidden object.

[0054] Next, the control device (100) of the autonomous vehicle determines a risk level based on the risk of collision between the autonomous vehicle and the hidden object (S312). Specifically, when the hidden object is set, the control device (100) of the autonomous vehicle can determine a risk level based on the driving information of the autonomous vehicle and the movement information of the hidden object. Additionally, when the hidden object is set in the identification module (150), the control device (100) of the autonomous vehicle can calculate the estimated time of arrival to the crosswalk based on the driving information of the autonomous vehicle and calculate the estimated time of crossing to the sidewalk based on the movement information of the hidden object. Additionally, the control device (100) of the autonomous vehicle can determine a risk level based on the risk of collision by comparing the estimated time of arrival and the estimated time of crossing.

[0055] Finally, the control device (100) of the autonomous vehicle controls the speed of the autonomous vehicle according to the risk level (S314).

[0056] Accordingly, according to embodiments of the present invention, the safety of an autonomous vehicle can be improved by sharing images of surrounding autonomous vehicles to identify objects located in blind spots.

[0058] FIG. 4 is a block diagram illustrating a computing environment including a computing device suitable for use in exemplary embodiments. In the illustrated embodiments, each component may have different functions and capabilities in addition to those described below, and may include additional components in addition to those described below.

[0059] The illustrated computing environment (10) includes a computing device (12). In one embodiment, the computing device (12) may be a control device (100) of an autonomous vehicle.

[0060] The computing device (12) includes at least one processor (14), a computer-readable storage medium (16), and a communication bus (18). The processor (14) can cause the computing device (12) to operate according to the exemplary embodiment described above. For example, the processor (14) can execute one or more programs stored in the computer-readable storage medium (16). The one or more programs may include one or more computer-executable instructions, and the computer-executable instructions may be configured to cause the computing device (12) to perform operations according to the exemplary embodiment when executed by the processor (14).

[0061] A computer-readable storage medium (16) is configured to store computer-executable instructions or program code, program data and / or other suitable forms of information. A program (20) stored in the computer-readable storage medium (16) includes a set of instructions executable by a processor (14). In one embodiment, the computer-readable storage medium (16) may be memory (volatile memory such as random access memory, non-volatile memory, or a suitable combination thereof), one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, or other forms of storage media that are accessed by a computing device (12) and capable of storing desired information, or a suitable combination thereof.

[0062] The communication bus (18) interconnects various other components of the computing device (12), including the processor (14) and the computer-readable storage medium (16).

[0063] The computing device (12) may also include one or more input / output interfaces (22) and one or more network communication interfaces (26) that provide interfaces for one or more input / output devices (24). The input / output interfaces (22) and network communication interfaces (26) are connected to a communication bus (18). The input / output devices (24) may be connected to other components of the computing device (12) through the input / output interfaces (22). An exemplary input / output device (24) may include an input device such as a pointing device (such as a mouse or trackpad), a keyboard, a touch input device (such as a touchpad or touchscreen), a voice or sound input device, various types of sensor devices and / or imaging devices, and / or an output device such as a display device, a printer, a speaker and / or a network card. An exemplary input / output device (24) may be included inside the computing device (12) as a component constituting the computing device (12), or it may be connected to the computing device (12) as a separate device distinct from the computing device (12).

[0065] Although representative embodiments of the present invention have been described in detail above, those skilled in the art will understand that various modifications can be made to the above-described embodiments without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined by the claims set forth below as well as equivalents thereof. Explanation of the symbols

[0067] 100: Control unit of an autonomous vehicle 110: Image sensor module 120: Communication module 130: Feature Extraction Module 140: Fusion Module 150 : Identification Module 160: Judgment Module 170: Driving control module

Claims

Claim 1 A control device for an autonomous vehicle comprising: an image sensor module for acquiring original image information captured from the front of an autonomous vehicle; a communication module for acquiring external image information from surrounding autonomous vehicles; a feature extraction module for extracting an original image frame from the original image information and extracting feature vector information for a plurality of objects within the original image frame, and for extracting an external image frame from the external image information and extracting feature vector information for a plurality of objects within the external image frame; a fusion module for fusing the original image frame and the external image frame based on feature vector information for a plurality of objects within the original image frame and feature vector information for a plurality of objects within the external image frame; an identification module for identifying a hidden object based on the fused image frame; a judgment module for determining a risk level based on the risk of collision between the identified hidden object and the autonomous vehicle; and a driving control module for controlling the speed of the autonomous vehicle according to the determined risk level, wherein the feature extraction module extracts feature vector information for a plurality of objects within the original image frame and a plurality of objects within the external image frame through SIFT (Scale-Invariant Feature Transform). Claim 2 delete Claim 3 A control device for an autonomous vehicle according to claim 1, wherein the fusion module determines the similarity between feature vector information for a plurality of objects of the original image frame and feature vector information for a plurality of objects of the external image frame, and performs matching between objects determined to have high similarity to fuse the original image frame and the external image frame. Claim 4 A control device for an autonomous vehicle according to claim 3, wherein the feature vector information includes the position of an object, the orientation of an object, the size of an object, and the gradient vector of an object, and the fusion module calculates the Euclidean distance between the gradient vectors for a plurality of objects in the original image frame and the gradient vectors for a plurality of objects in the external image frame, and determines that objects with a distance less than or equal to a preset distance have high similarity. Claim 5 A control device for an autonomous vehicle comprising: an image sensor module for acquiring original image information captured of the front of an autonomous vehicle; a communication module for acquiring external image information from surrounding autonomous vehicles; a feature extraction module for extracting an original image frame from the original image information and extracting feature vector information for a plurality of objects within the original image frame, and for extracting an external image frame from the external image information and extracting feature vector information for a plurality of objects within the external image frame; a fusion module for fusing the original image frame and the external image frame based on feature vector information for a plurality of objects within the original image frame and feature vector information for a plurality of objects within the external image frame; an identification module for identifying a hidden object based on the fused image frame; a judgment module for determining a risk level based on the risk of collision between the identified hidden object and the autonomous vehicle; and a driving control module for controlling the speed of the autonomous vehicle according to the determined risk level, wherein the identification module detects an object existing in one image frame by comparing the fused image frame and the original image frame, and identifies the detected object and sets it as a hidden object. Claim 6 In claim 5, the identification module identifies the detected object by an object recognition method based on Faster R-CNN (Regions with Convolutional Neuron Networks), a control device for an autonomous vehicle. Claim 7 An image sensor module for acquiring original image information captured from the front of an autonomous vehicle; a communication module for acquiring external image information from surrounding autonomous vehicles; a feature extraction module for extracting an original image frame from the original image information and extracting feature vector information for a plurality of objects within the original image frame, and for extracting an external image frame from the external image information and extracting feature vector information for a plurality of objects within the external image frame; a fusion module for fusing the original image frame and the external image frame based on the feature vector information for a plurality of objects within the original image frame and the feature vector information for a plurality of objects within the external image frame; an identification module for identifying a hidden object based on the fused image frame; and a judgment module for determining a risk level based on the risk of collision between the identified hidden object and the autonomous vehicle. A control device for an autonomous vehicle, comprising a driving control module that controls the speed of the autonomous vehicle according to the determined risk level, wherein the feature extraction module acquires the original image information and the external image information based on previously stored road information and original vehicle information acquired from the autonomous vehicle, and when the distance between the autonomous vehicle and the crosswalk is within a preset distance, the external image information is acquired from another autonomous vehicle within a preset distance in the direction of travel of the autonomous vehicle. Claim 8 A control device for an autonomous vehicle according to claim 7, wherein the judgment module calculates the estimated time of arrival to the crosswalk based on original vehicle information obtained from the autonomous vehicle, calculates the estimated time of crossing to the sidewalk based on the movement information of the hidden object, determines the risk level by comparing the estimated time of arrival and the estimated time of crossing, and the movement information of the hidden object is generated by extracting the position, speed, and direction of travel of the hidden object from consecutive frames of the fused image frames. Claim 9 A method for controlling an autonomous vehicle, performed on a computing device having one or more processors and a memory for storing one or more programs executed by said one or more processors, comprising: acquiring original image information capturing the front of the autonomous vehicle and acquiring external image information from surrounding autonomous vehicles; extracting an original image frame from said original image information and extracting feature vector information for a plurality of objects within said original image frame, respectively; extracting an external image frame from said external image information and extracting feature vector information for a plurality of objects within said external image frame, respectively; fusing said original image frame and said external image frame based on the feature vector information for a plurality of objects within said original image frame and the feature vector information for a plurality of objects within said external image frame; identifying a hidden object based on said fused image frame; and determining a risk level according to the risk of collision between said identified hidden object and said autonomous vehicle. A method for controlling an autonomous vehicle, comprising a driving control module that controls the speed of the autonomous vehicle according to the determined risk level, wherein the extraction step further comprises the step of extracting feature vector information for a plurality of objects within the original image frame and a plurality of objects within the external image frame through SIFT (Scale-Invariant Feature Transform). Claim 10 delete Claim 11 A control method for an autonomous vehicle according to claim 9, wherein the fusing step further comprises: a step of determining similarity between feature vector information for a plurality of objects of the original image frame and feature vector information for a plurality of objects of the external image frame; and a step of fusing the original image frame and the external image frame by performing matching between objects determined to have high similarity. Claim 12 A control method for an autonomous vehicle according to claim 11, wherein the feature vector information includes the position of an object, the orientation of an object, the size of an object, and the gradient vector of an object, and the fusing step further comprises: a step of calculating the Euclidean distance between the gradient vectors for a plurality of objects in the original image frame and the gradient vectors for a plurality of objects in the external image frame; and a step of determining that the similarity is high for objects that are less than or equal to a preset distance as a result of the calculation. Claim 13 A method for controlling an autonomous vehicle, performed on a computing device having one or more processors and a memory for storing one or more programs executed by said one or more processors, comprising: acquiring original image information capturing the front of the autonomous vehicle and acquiring external image information from surrounding autonomous vehicles; extracting an original image frame from said original image information and extracting feature vector information for a plurality of objects within said original image frame, respectively; extracting an external image frame from said external image information and extracting feature vector information for a plurality of objects within said external image frame, respectively; fusing said original image frame and said external image frame based on the feature vector information for a plurality of objects within said original image frame and the feature vector information for a plurality of objects within said external image frame; identifying a hidden object based on said fused image frame; and determining a risk level according to the risk of collision between said identified hidden object and said autonomous vehicle. A method for controlling an autonomous vehicle, comprising a driving control module that controls the speed of the autonomous vehicle according to the determined risk level, wherein the identifying step further comprises: a step of detecting an object present in one image frame by comparing the fused image frame and the original image frame; and a step of identifying the detected object and setting it as a hidden object. Claim 14 A control method for an autonomous vehicle according to claim 13, wherein the identifying step identifies the detected object by an object recognition method based on Faster R-CNN (Regions with Convolutional Neuron Networks). Claim 15 A method for controlling an autonomous vehicle, performed on a computing device having one or more processors and a memory for storing one or more programs executed by said one or more processors, comprising: acquiring original image information capturing the front of the autonomous vehicle and acquiring external image information from surrounding autonomous vehicles; extracting an original image frame from said original image information and extracting feature vector information for a plurality of objects within said original image frame, respectively; extracting an external image frame from said external image information and extracting feature vector information for a plurality of objects within said external image frame, respectively; fusing said original image frame and said external image frame based on the feature vector information for a plurality of objects within said original image frame and the feature vector information for a plurality of objects within said external image frame; identifying a hidden object based on said fused image frame; and determining a risk level according to the risk of collision between said identified hidden object and said autonomous vehicle. A method for controlling an autonomous vehicle, comprising a driving control module that controls the speed of the autonomous vehicle according to the determined risk level, wherein the extraction step further comprises: a step of acquiring the original image information and the external image information when the distance between the autonomous vehicle and the crosswalk is within a preset distance based on previously stored road information and original vehicle information acquired from the autonomous vehicle; and a step of acquiring the external image information from another autonomous vehicle within a preset distance in the direction of travel of the autonomous vehicle. Claim 16 A method for controlling an autonomous vehicle according to claim 15, wherein the determining step further comprises: a step of calculating the estimated time of arrival to the crosswalk based on original vehicle information obtained from the autonomous vehicle; a step of calculating the estimated time of crossing to the sidewalk based on the movement information of the hidden object; and a step of determining the risk level by comparing the estimated time of arrival and the estimated time of crossing, wherein the movement information of the hidden object is generated by extracting the position, speed, and direction of travel of the hidden object from consecutive frames of the fused image frames.

Images