KR-20260065046-A - Method And Apparatus for Object Matching And Calibration between Autonomous Vehicle And Infrastructure

Abstract

An object matching and correction device between an autonomous vehicle and infrastructure is disclosed. According to one aspect of the present disclosure, an object matching and correction device between an autonomous vehicle and an infrastructure comprises: a first sensor module installed in the autonomous vehicle for identifying an object within a sensing range and detecting vehicle object data; a second sensor module installed in the infrastructure for identifying an object within a sensing range and detecting infrastructure object data; a communication module installed in the autonomous vehicle and the infrastructure, respectively, for transmitting object data in real time between the autonomous vehicle and the infrastructure using wireless communication; and a data processing module installed in the autonomous vehicle for determining whether the infrastructure object data received from the second sensor module is valid by comparing the communication delay time with an allowable communication delay time calculated using interpolation on the speed of the autonomous vehicle, and if it is determined to be valid, correcting the communication delay of the infrastructure object data, tracking the communication delay-corrected infrastructure object data and the vehicle object data using an object tracking algorithm, and performing matching using the object-tracked infrastructure object data and the object-tracked vehicle object data.

Inventors

• 성재호
• 김태형
• 양은주
• 김봉섭
• 윤경수

Assignees

• 재단법인 지능형자동차부품진흥원

Dates

Publication Date

20260508

Application Date

20241031

Claims (20)

1. In an object matching and correction device between an autonomous vehicle and infrastructure, A first sensor module installed in the above-mentioned autonomous vehicle for identifying objects within a sensing range and detecting vehicle object data; A second sensor module installed in the above infrastructure to identify objects within a sensing range and detect infrastructure object data; A communication module installed in each of the autonomous vehicle and the infrastructure, and for transmitting object data in real time between the autonomous vehicle and the infrastructure using wireless communication; and A device comprising a data processing module for installing on the autonomous vehicle, determining whether the infrastructure object data received from the second sensor module is valid by comparing the communication delay time with an allowable communication delay time calculated using interpolation on the speed of the autonomous vehicle, and if it is determined to be valid, correcting the communication delay of the infrastructure object data, tracking the communication delay-corrected infrastructure object data and the vehicle object data using an object tracking algorithm, and performing alignment using the object-tracked infrastructure object data and the object-tracked vehicle object data.
2. In Article 1, The above infrastructure object data comprises at least one of an object identification number, object reliability, object appearance, and object movement information, in a device.
3. In Article 1, The above data processing module is a device that deletes the infrastructure object data when it is determined that the infrastructure object data received from the second sensor module is invalid.
4. In Article 1, The above object tracking algorithm is a device that continuously tracks and synchronizes object state information for time points at the detection time intervals of the vehicle object data and the infrastructure object data.
5. In Article 1, The above matching is a device that removes duplicate objects and generates a final object list using the object-tracked infrastructure object data and the object-tracked vehicle object data.
6. In Article 1, In the above data processing module performing the above matching, Using the center point of the vehicle object extracted from the above object-tracked vehicle object data as a reference point, the value obtained by dividing the length of the vehicle object of the 3D bounding box by 2 is determined as the radius, and The infrastructure object distance is calculated using the Euclidean distance between the center point of the infrastructure object extracted from the object-tracked infrastructure object data and the center point of the vehicle object extracted from the object-tracked vehicle object data. A device that generates a final object list containing different types of object data based on whether there exists an infrastructure object whose distance is smaller than or equal to the radius.
7. In Article 6, In the above data processing module generating the final object list, A device that adds vehicle object data of a vehicle object that served as the reference point of the radius to the final object list when the infrastructure object distance is smaller than or equal to the radius and the infrastructure object does not exist.
8. In Article 6, In generating the final object list, the above data processing module If there are multiple first infrastructure objects whose distance from the above infrastructure object is smaller than or equal to the above radius, one first infrastructure object is selected from among the multiple first infrastructure objects, and A device that adds infrastructure object data of the first selected infrastructure object to the final object list.
9. In Article 8, The above addition is, A device that, if there is a second infrastructure object whose distance from the infrastructure object is greater than the radius, aligns the infrastructure object data of the second infrastructure object with the infrastructure object data of the selected first infrastructure object and adds it to the final object list.
10. In Article 8, Selecting the above-mentioned first infrastructure object is, A device that calculates a duplication judgment score indicating the degree of duplication between a vehicle object and a first infrastructure object, and selects the infrastructure object with the lowest duplication judgment score.
11. In a method for object matching and correction between an autonomous vehicle and infrastructure, A process of identifying objects within the sensing range and acquiring infrastructure object data; A process of identifying objects within a sensing range and acquiring vehicle object data; A process of transmitting object data in real time between the autonomous vehicle and the infrastructure using wireless communication; A process of determining whether the infrastructure object data received from the second sensor module is valid by comparing the communication delay time with an allowable communication delay time calculated using interpolation on the speed of the autonomous vehicle - the second sensor module is installed in the infrastructure, identifies objects within the sensing range, and detects infrastructure object data -; If the above infrastructure object data is determined to be valid, a process of correcting the communication delay of the above infrastructure object data; A process of object tracking using an object tracking algorithm with infrastructure object data corrected for communication delay and the vehicle object data; and A method comprising a process of performing alignment using object-tracked infrastructure object data and object-tracked vehicle object data.
12. In Article 11, The process of determining whether the above validity exists is, If the above communication delay time is greater than the above allowable communication delay time, the above infrastructure object data is determined to be invalid, and If the above communication delay time is less than or equal to the above allowable communication delay time, the infrastructure object data is determined to be valid, but The above communication delay time is, It is calculated by subtracting the infrastructure object detection time from the total latency, and The above allowable communication delay time is, A method for calculating the speed of the above-mentioned autonomous vehicle using interpolation.
13. In Article 12, A method for deleting infrastructure object data when it is determined that the infrastructure object data is invalid.
14. In Article 11, The above process of correcting communication delay is, The process of calculating the estimated value of the previous state; A process of predicting a predicted value of the current state based on an estimated value of the previous state; and A method comprising a process of updating the predicted value of the current state based on the above infrastructure object data.
15. In Article 11, The above object tracking process is, A method for continuously tracking and synchronizing object state information for time points between the detection time points of the communication delay-corrected infrastructure object data and the vehicle object data using the above object tracking algorithm.
16. In Article 11, The process of performing the above matching is, A process of determining the radius by dividing the length of the vehicle object of the 3D bounding box by 2, using the center point of the vehicle object extracted from the above-mentioned object-tracked vehicle object data as a reference point; A process of calculating an infrastructure object distance using the Euclidean distance between the center point of an infrastructure object extracted from the object-tracked infrastructure object data and the center point of a vehicle object extracted from the object-tracked vehicle object data; and A method comprising the process of generating a final object list containing different types of object data based on whether there exists an infrastructure object whose distance from the infrastructure object is smaller than or equal to the radius.
17. In Article 16, The process of generating the above final object list is, A method comprising the process of adding vehicle object data of the vehicle object that served as the reference point of the radius to the final object list when the infrastructure object distance is smaller than or equal to the radius and the infrastructure object does not exist.
18. In Article 16, The process of generating the above final object list is, When there are multiple first infrastructure objects whose distance from the above infrastructure object is smaller than or equal to the radius, the process of selecting one first infrastructure object among the multiple first infrastructure objects; and A method comprising the process of adding infrastructure object data of the first selected infrastructure object to the final object list.
19. In Article 18, The above adding process is, A method in which, if there exists a second infrastructure object whose distance from the infrastructure object is greater than the radius, the infrastructure object data of the second infrastructure object and the infrastructure object data of the selected first infrastructure object are matched and added to the final object list.
20. In Article 18, The process of selecting the above-mentioned first infrastructure object is, A process of calculating a duplication judgment score indicating the degree of duplication between the above vehicle object and the above first infrastructure object; A method comprising the process of selecting the infrastructure object with the lowest duplicate judgment score.

Description

Method and Apparatus for Object Matching and Calibration between Autonomous Vehicle and Infrastructure The present disclosure relates to an apparatus and method for object alignment and correction between an autonomous vehicle and infrastructure. Specifically, the present disclosure relates to an apparatus and method for object alignment and correction between an autonomous vehicle and infrastructure that shares multiple object data collected through infrastructure and an autonomous vehicle in real time, determines data validity, corrects communication delay, and performs object tracking and alignment. The following description merely provides background information related to the present embodiment and does not constitute prior art. Autonomous vehicles refer to automobiles capable of operating independently without human intervention. Driver assistance systems are essential for detecting surrounding hazardous objects through onboard sensors and warning the driver to respond to dangerous situations. Sensors used to detect surrounding objects include LiDAR, radar, and cameras, which recognize various indicators such as the movement and appearance of objects. Previously, vehicles or infrastructure recognized and detected objects in three dimensions using their own sensors without mutual cooperation. While the existing method did not have transmission delay issues because it did not rely on communication, such as data transmission and reception between vehicles and infrastructure, it had limitations in recognizing the surrounding environment. For instance, there were several limitations, such as the recognition range being restricted to the sensing range capable of object detection, or a decrease in the accuracy of identifying the surrounding environment. There is a method for each system to share object data detected by the other. This approach allows for the sharing of data from other systems even regarding situations occurring outside the system's sensing range, thereby improving the overall safety of autonomous driving. However, there are drawbacks to the method of performing cooperative object detection between vehicles and infrastructure using object data sharing. For example, there are issues such as communication delays, differences in object detection cycles, duplicate objects, and a lack of synchronization mechanisms. The communication delay issue is that delays in communication between the vehicle and the infrastructure can cause object data detected by the infrastructure to lose validity in the autonomous driving system. The object detection cycle difference problem is an issue where unsynchronized data occurs for the same object when vehicles and infrastructure detect objects at different cycles. The duplicate object problem is that duplicate data can occur when the same object is detected in vehicles and infrastructure, and there is a lack of efficient methods to handle this. The problem of insufficient synchronization mechanisms is that data alignment is difficult due to the lack of synchronization and correction algorithms capable of resolving differences in communication delays and object detection cycles. FIG. 1 is an illustrative diagram for explaining an object matching and correction device according to one embodiment of the present disclosure. FIG. 2 is a block diagram schematically showing an object matching and correction device according to one embodiment of the present disclosure. FIG. 3 is a flowchart illustrating an object matching and correction method according to one embodiment of the present disclosure. FIG. 4 is a flowchart illustrating an object matching and correction method according to one embodiment of the present disclosure. FIG. 5 is an illustrative diagram for explaining a three-dimensional bounding box in an object tracking process and a matching process according to one embodiment of the present disclosure. FIG. 6 is a flowchart illustrating an object alignment and correction method according to one embodiment of the present disclosure. Some embodiments of the present disclosure are described in detail below with reference to exemplary drawings. It should be noted that in assigning reference numerals to the components of each drawing, the same components are given the same reference numeral whenever possible, even if they are shown in different drawings. Furthermore, in describing the present disclosure, if it is determined that a detailed description of related known components or functions could obscure the essence of the present disclosure, such detailed description is omitted. In describing the components of the embodiments according to the present disclosure, symbols such as first, second, i), ii), a), b), etc., may be used. These symbols are intended only to distinguish the components from other components, and the essence, order, or sequence of the components is not limited by the symbols. When a part in the specification is described as 'comprising' or 'having