Method for generating three-dimensional driving perception data, computer program product, driver assistance system and motor vehicle

The method improves autonomous driving by integrating parallel image and aerial processing paths with LiDAR data to create a reliable three-dimensional environmental map, addressing object detection challenges and enhancing safety in complex environments.

DE102024211397A1Pending Publication Date: 2026-05-28FCA US LLC +1
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
DE102024211397
Authority / Receiving Office
DE · DE
Patent Type
Applications
Current Assignee / Owner
Filing Date
2024-11-28
Publication Date
2026-05-28

AI Technical Summary

Technical Problem

Existing autonomous driving systems face challenges in ensuring reliable and accurate detection and classification of objects in complex, dynamic environments, particularly under adverse conditions, which can lead to misclassifications and safety risks due to impaired sensors and obscured objects.

Method used

A method for generating three-dimensional driving perception data using parallel image processing paths that combine safety attribute data from a camera with depth data from an aerial image generation path, incorporating perception heads and BEV encoders to enhance object detection and classification, and integrating LiDAR data for precise depth estimation.

Benefits of technology

This approach provides a comprehensive and reliable environmental map for autonomous vehicles, enhancing object detection accuracy and safety by leveraging the strengths of both image and aerial processing paths, and ensuring precise and context-dependent decision-making.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

A method for generating three-dimensional driving perception data for an automated driving system of a motor vehicle is described, wherein image data of a traffic situation is recorded using at least one camera provided in the motor vehicle, wherein the image data is processed in parallel in an image processing path and in an aerial image generation path, wherein in the image processing path first object data with safety attribute data as well as pseudo-depth data are generated based on the image data, wherein a bird's-eye view representation is generated in the aerial image generation path, wherein second object data with depth data are generated in the aerial image generation path, wherein the data from the image processing path and from the aerial image generation path are merged, wherein the safety attribute data from the image processing path are assigned to the second object data.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Three dimensional (3D) object detection

    US12051213B2

  • Object detection and detection confidence suitable for autonomous driving

    US20240192320A1

  • CROSS-MODALITY ACTIVE LEARNING FOR OBJECT RECOGNITION

    DE102021133340A1