CN-122024487-A - Automobile safety perception early warning system

CN122024487ACN 122024487 ACN122024487 ACN 122024487ACN-122024487-A

Abstract

The invention provides an automobile safety perception early warning system which comprises a physical signal lamp perception module, a network communication module, a high-precision time synchronization module and a data fusion module. The system comprises a physical signal lamp sensing module, a network communication module, a high-precision time synchronization module, a data fusion module and a vehicle, wherein the physical signal lamp sensing module collects real-time images and recognizes the real-time images to generate first sensing data containing time stamps, the network communication module receives a road side unit V2X message and analyzes second sensing data containing digital signal lamp states and the time stamps, the high-precision time synchronization module is connected with a satellite time service system to provide a unified time reference for the two signals and ensure alignment of the data time stamps, and the data fusion module fuses the two types of data based on the alignment time stamps to generate accurate signal lamp state information and provide safety pre-warning for the vehicle. According to the invention, through high-precision time synchronization and multi-source data fusion, the advantages of visual perception and network communication are effectively combined, the accuracy and reliability of intelligent network-connected automobile in traffic signal perception under complex traffic environment are obviously improved, and further the driving safety is enhanced.

Inventors

NIU CHAO
WU HAO
LI ZELONG
TAO YUNFEI
CHEN ZHUO
ZHENG HONG
GAO YUDE
ZHANG KENIU
Wang Yongchuang
ZHANG YUNLONG

Assignees

长春汽车检测中心有限责任公司

Dates

Publication Date: 20260512
Application Date: 20260324

Claims (10)

1. An automotive safety perception early warning system, the system comprising: The physical signal lamp sensing module is used for acquiring a real-time image of a physical signal lamp and identifying the image to generate first sensing data containing signal lamp state information and acquisition time stamps thereof; The network communication module is used for receiving the V2X message broadcast by the road side unit and analyzing second perception data containing the digital signal lamp state information and the timestamp thereof from the V2X message; The high-precision time synchronization module is connected with the satellite time service system and is respectively connected with the physical signal lamp sensing module and the network communication module, and is used for providing a unified time reference for the physical signal lamp sensing module and the network communication module, so that the time stamp of the first sensing data is aligned with the time stamp of the second sensing data in time; The data fusion module is respectively connected with the physical signal lamp sensing module, the network communication module and the high-precision time synchronization module, and is used for carrying out fusion processing on the first sensing data and the second sensing data based on the aligned time stamp, generating fused signal lamp state information and providing safety early warning for a vehicle based on the fused signal lamp state information.
2. The automotive safety awareness early warning system of claim 1, wherein the physical signal light awareness module comprises: The image acquisition unit is used for acquiring images of the physical signal lamp at a frequency not lower than a preset frame rate and recording the exposure starting time of each frame of image as the acquisition time stamp; And the image recognition unit is used for recognizing the acquired image through a deep learning algorithm to obtain the signal lamp state information, wherein the signal lamp state information at least comprises a lamp color state and a countdown numerical value.
3. The vehicle safety perception early warning system according to claim 2, wherein the image acquisition unit employs a global shutter industrial camera, and the image recognition unit is provided with an AI acceleration chip, and is configured to operate a semantic segmentation network to locate signal light areas in an image, and to operate a classification network to recognize light color states and countdown values in the areas.
4. The vehicle safety awareness early warning system of claim 1, wherein the V2X message received by the web communication module comprises at least a SPAT message and a MAP message compliant with a standard protocol, and the second awareness data is parsed from the SPAT message and comprises at least a signal lamp phase ID, a lamp color status, and timing information.
5. The automotive safety-aware early warning system of claim 1, wherein the high-precision time synchronization module follows a precision time protocol PTP that provides a timing precision such that a synchronization error between a timestamp of the first perceived data and a timestamp of the second perceived data is less than 1 millisecond.
6. The vehicle safety awareness early warning system of claim 1, wherein the data fusion module comprises: the data alignment unit is used for carrying out time registration on the first sensing data and the second sensing data according to a unified time reference; the consistency check unit is used for comparing the consistency of the signal lamp state information in the first sensing data and the second sensing data at the same time and generating an alarm signal when the signal lamp state information is inconsistent; And the fusion output unit is used for fusing the first perception data and the second perception data according to a preset fusion strategy and outputting signal lamp state information with higher reliability.
7. The system of claim 6, wherein the consistency check unit is further configured to calculate a time difference between the changes in the color states of the lamps in the first sensing data and the second sensing data, and determine that the delay of the internet message is abnormal if the time difference exceeds a preset threshold.
8. The vehicle safety perception early warning system according to claim 1, wherein the data fusion module is further configured to determine a phase ID of a signal lamp corresponding to a lane where the vehicle is currently located according to the MAP message in the second perception data, and associate the phase ID with the signal lamp identified in the first perception data.
9. The vehicle safety awareness early warning system of claim 1 wherein the physical signal light awareness module further comprises a human-machine interface for a user to manually associate or modify signal light phase IDs.
10. The vehicle safety perception early warning system according to claim 1, further comprising a vehicle-mounted display unit for displaying the fused signal lamp state information to a driver, and sending out a visual or audible early warning when the first perception data and the second perception data are detected to be inconsistent or the network connection message is absent.

Description

Automobile safety perception early warning system Technical Field The invention relates to the technical field of automobile safety perception, in particular to an automobile safety perception early warning system. Background At present, the intelligent network-connected automobile mainly has two approaches to the perception of traffic signals, namely, firstly, a physical signal lamp image is collected and identified through an on-board camera, and secondly, a V2X message broadcast by a Road Side Unit (RSU) is received through an on-board unit (OBU), such as a signal lamp state message (SPAT) and a MAP Message (MAP) which follow standard protocols, so as to acquire digital lamp color, countdown and phase information. However, both approaches have limitations when used alone. Visual perception is susceptible to interference from factors such as illumination changes (e.g., backlight, overexposure), bad weather (e.g., rain, snow, fog), and physical shielding (e.g., large vehicles, trees), resulting in reduced recognition stability. The network communication mode may face the problems of signal delay, packet loss, incomplete message content or inconsistent formats, and the like, so as to affect the real-time performance and reliability of the data. In the prior art, a system capable of effectively fusing visual perception data and network communication data and performing high-precision time alignment and consistency verification on the visual perception data and the network communication data is not available. Due to the lack of a unified time reference, two types of data are difficult to match accurately in the time dimension, so that the fusion effect is poor, continuous and accurate signal lamp state information cannot be provided for a vehicle in a complex scene, and the effectiveness of the intelligent network-connected automobile safety early warning system is restricted. Therefore, how to realize the deep fusion of the vision and the internet-connected information so as to improve the robustness and the accuracy of the traffic signal perception becomes a technical problem to be solved currently. Disclosure of Invention The invention aims to provide an automobile safety perception early warning system for solving the problems in the background technology. The invention is realized by the following technical scheme: An automotive safety awareness early warning system, the system comprising: The physical signal lamp sensing module is used for acquiring a real-time image of a physical signal lamp and identifying the image to generate first sensing data containing signal lamp state information and acquisition time stamps thereof; The network communication module is used for receiving the V2X message broadcast by the road side unit and analyzing second perception data containing the digital signal lamp state information and the timestamp thereof from the V2X message; The high-precision time synchronization module is connected with the satellite time service system and is respectively connected with the physical signal lamp sensing module and the network communication module, and is used for providing a unified time reference for the physical signal lamp sensing module and the network communication module, so that the time stamp of the first sensing data is aligned with the time stamp of the second sensing data in time; The data fusion module is respectively connected with the physical signal lamp sensing module, the network communication module and the high-precision time synchronization module, and is used for carrying out fusion processing on the first sensing data and the second sensing data based on the aligned time stamp, generating fused signal lamp state information and providing safety early warning for a vehicle based on the fused signal lamp state information. Further, the physical signal lamp sensing module includes: The image acquisition unit is used for acquiring images of the physical signal lamp at a frequency not lower than a preset frame rate and recording the exposure starting time of each frame of image as the acquisition time stamp; And the image recognition unit is used for recognizing the acquired image through a deep learning algorithm to obtain the signal lamp state information, wherein the signal lamp state information at least comprises a lamp color state and a countdown numerical value. Furthermore, the image acquisition unit adopts a global shutter industrial camera, and the image recognition unit is internally provided with an AI acceleration chip and is used for operating a semantic segmentation network to locate signal lamp areas in the image and operating a classification network to recognize lamp color states and countdown numerical values in the areas. Further, the V2X message received by the network communication module at least includes a SPAT message and a MAP message that conform to a standard protocol, and the second sensing data is obtained by parsing the SPAT message,