KR-20260063817-A - Apparatus recognizing a jaywalker for autonomous cooperative driving and method using the same

Abstract

The present invention relates to a jaywalking pedestrian recognition device for generating a precision map for autonomous cooperative driving and a method using the same, characterized by comprising: an image collection unit that collects image data captured from a camera provided in the infrastructure; an object recognition unit that recognizes road boundaries and lanes constituting the road and objects located on the road using an artificial intelligence model from the image data; a jaywalking pedestrian recognition unit that determines whether a pedestrian is jaywalking using the recognized object information and sets the location of the accident area and the accident section; a map generation unit that maps the jaywalking pedestrian to a precision map; and a map transmission unit that transmits the precision map to a map server.

Inventors

• 오광만
• 추연호
• 안웅

Assignees

• (주)테슬라시스템

Dates

Publication Date

20260507

Application Date

20241031

Claims (5)

1. In a pedestrian crossing recognition device that generates a precision map for autonomous cooperative driving, A video collection unit that collects video data captured from a camera equipped in the infrastructure; An object recognition unit that recognizes road boundaries and lanes constituting the road and objects located on the road using an artificial intelligence model from the above image data; A jaywalking pedestrian recognition unit that determines whether a pedestrian is jaywalking by utilizing the above-mentioned recognized object information and predicts the location and direction of movement where the jaywalking is taking place; A map generation unit that maps the above-mentioned jaywalking pedestrian onto the above-mentioned precision map; and A pedestrian jaywalking recognition device characterized by comprising a map transmission unit that transmits the above-mentioned precision map to a map server.
2. In paragraph 1, The above object recognition unit is, A road recognition unit that recognizes road boundaries forming a road and lanes within the road to recognize the above-mentioned jaywalking pedestrian; A pedestrian recognition unit that recognizes a pedestrian walking on the aforementioned road; A walking direction determination unit that determines the direction of travel by tracking the location of the recognized pedestrian; and A pedestrian jaywalking recognition device characterized by comprising a coordinate conversion unit that converts local coordinates within the image into global coordinates to reflect the above location in the above precision map.
3. In paragraph 1, The above-mentioned jaywalking pedestrian recognition unit includes a vector cross product calculation unit that calculates a vector cross product using a direction of travel vector and a walking direction vector; and A jaywalking pedestrian recognition device characterized by comprising a jaywalking pedestrian determination unit that selects the jaywalking pedestrian using the above vector cross product.
4. In paragraph 1, A pedestrian jaywalking recognition device characterized by the above map transmission unit transmitting the above precision map, including the jaywalking pedestrian, in real time to an autonomous vehicle located within the coverage range of the above infrastructure using V2X communication.
5. In a method of using a pedestrian jaywalking recognition device according to any one of paragraphs 1 to 3, A first step of recognizing road boundaries using a camera equipped in the infrastructure to set a road area; A second step of recognizing pedestrians walking among objects located within the above road area and tracking these pedestrians; The third stage of recognizing pedestrians within the road area other than the crosswalk as jaywalkers and A method for recognizing jaywalking pedestrians characterized by comprising a fourth step of mapping the aforementioned jaywalking pedestrian onto a precision map.

Description

Apparatus recognizing a jaywalker for autonomous cooperative driving and method using the same The present invention relates to a pedestrian jaywalking recognition device for autonomous cooperative driving and a method using the same, which recognizes pedestrians jaywalking on a road using a camera installed in infrastructure along a roadside and reflects this in real time on a precision map for an autonomous vehicle. An autonomous vehicle (hereinafter referred to as "autonomous vehicle") has the function of recognizing environmental information necessary for driving, such as the types and locations of objects around the vehicle, using various sensors such as radar and lidar and cameras, and controlling the vehicle based on this information to perform autonomous driving. However, along the routes autonomous vehicles intend to travel, there are always situations where they cannot respond in real time due to traffic accidents, unexpected emergencies, and blind spots. As such, an edge infrastructure system (hereinafter referred to as "infrastructure") equipped with various sensors installed along the roadside is required to recognize the autonomous driving environment of the autonomous vehicle in advance. In addition to the function of detecting moving objects around the road (e.g., people, cars, bicycles, etc.), this infrastructure needs to recognize information on unexpected situations, road surface conditions, traffic accidents, and road accidents, and reflect this in a precision map for autonomous driving. Precision maps for autonomous vehicles can be displayed by classifying them into static elements of a typical road (roadway, lane markings, road signs, traffic light locations, traffic signs, etc.) and dynamic elements that change over time (obstacles, accident zones, traffic congestion, surrounding pedestrians, pedestrians, etc.). While the purpose of these precision maps is no different from current vehicle navigation maps, they require more accurate location information for static elements and a rapid update cycle for dynamic elements; additionally, they can safely perform autonomous driving functions by selecting the optimal route from the starting point to the destination. In particular, when a pedestrian jaywalks on a road detectable by infrastructure, the autonomous vehicle may find itself in an unavoidable situation by the time it detects them. Therefore, it is necessary to recognize this in advance, reflect it in a precision map, and provide a driving route that can detour or avoid the situation. However, conventional infrastructure lacks the means to recognize jaywalking pedestrians, and even if it does recognize them, it fails to provide the autonomous vehicle with information regarding which lane of the road and at what speed the pedestrian is moving. Patent Document 1 relates to recognizing pedestrians violating traffic laws by detecting the position and speed of pedestrians walking on a road using multiple cameras, and describes a means of recognizing pedestrians using cameras installed in each lane to check whether they are jaywalking. However, Patent Document 1 has the problem that cameras must be installed on every lane, which increases the cost of installing the infrastructure. FIG. 1 is a configuration diagram showing the configuration of an infrastructure including a pedestrian jaywalking recognition device according to an embodiment of the present invention. FIG. 2 is a diagram showing an edge computer that drives a pedestrian crossing illegally according to an embodiment of the present invention. FIG. 3 is a functional block diagram of a pedestrian jaywalking recognition device according to an embodiment of the present invention. FIG. 4 is a diagram showing a functional block diagram of an object recognition unit according to an embodiment of the present invention. FIG. 5 is a drawing showing an example of recognizing a road boundary and a pedestrian walking on the road according to an embodiment of the present invention. FIG. 6 is a diagram illustrating the direction of travel of a recognized pedestrian according to an embodiment of the present invention. FIG. 7 is a functional block diagram of a pedestrian jaywalking recognition unit according to an embodiment of the present invention. FIG. 8 is an example drawing for determining whether a pedestrian is jaywalking according to an embodiment of the present invention. FIG. 9 illustrates an example of converting the position of an image acquired from an image camera attached to an infrastructure according to an embodiment of the present invention into global coordinates. FIG. 10 is a flowchart illustrating a method of using a pedestrian jaywalking recognition device according to an embodiment of the present invention. Embodiments that enable a person skilled in the art to easily implement the present invention are described in detail below with reference to the attached drawings. However, in describing the operating principles of pre