Road Model Generation via Multi-Source Lane Line Arbitration

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing road model construction methods for autonomous driving systems face challenges in maintaining accuracy and stability, particularly in complex scenarios like rain or strong light, due to the limitations of relying on a single source of lane line information, such as high-definition maps or visual detection.

Innovation Solution

A method that comprehensively uses lane line information from multiple sources, including high-definition maps and visually detected lane lines, through arbitration and selection to determine the most accurate lane lines, ensuring that the selected lane lines match the reference lane lines and adhere to conditions of maximum and minimum lane widths, thereby improving the stability and accuracy of the road model.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If visual detection is used to obtain lane line information, then real-time performance is improved, but accuracy deteriorates in complex scenarios such as rain or strong light

Engineering Contradiction:
Improvereal-time performanceVSAvoidlane line detection accuracy
Core Design Contradiction:
SpeedVSMeasurement precision

Solution Approach 1:

The patent combines multiple sources of lane line information including visual detection results and high-definition map data to compensate for the weaknesses of single-source detection. By merging these sources through arbitration mechanisms, the system maintains real-time performance while improving accuracy in complex scenarios where visual detection alone fails.

Inventive Principle:
Principle #5Merging (Combining)

2Measurement precision

If high-definition map information is used to obtain lane line information, then accuracy is improved, but real-time performance deteriorates due to positioning dependency and update frequency

Engineering Contradiction:
Improvelane line information accuracyVSAvoidinformation update speed
Core Design Contradiction:
Measurement precisionVSSpeed

Solution Approach 1:

The patent implements a dynamic arbitration mechanism that adaptively selects between high-definition map information and visual detection results based on current driving conditions, scenario complexity, and data freshness. This dynamic approach allows the system to leverage the high accuracy of map data when available and reliable, while switching to real-time visual detection when map data is outdated or unavailable, thus balancing accuracy and real-time performance.

Inventive Principle:
Principle #15Dynamics

3Device complexity

If single source lane line information is used, then device complexity is reduced, but reliability deteriorates in complex scenarios

Engineering Contradiction:
Improveinformation processing complexityVSAvoidroad model stability
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent introduces an arbitration module as an intermediary component that coordinates between multiple lane line information sources. This mediator evaluates the reliability of each source based on current conditions, resolves conflicts between conflicting information, and selects the most reliable lane line data for road model construction, thereby improving reliability without requiring complete reprocessing of all source data.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4125061A1Road model generation method and device
Publication Date: 2023.02.01 NIO TECH ANHUI CO LTD
  • EP4125061A1 patent drawingFigure 1~2
  • EP4125061A1 patent drawingFigure 3
  • EP4125061A1 patent drawingFigure 4

AI summary

The disclosure relates to a road model generation method, the method including: receiving lane line information from a plurality of sources; determining a reference lane line; and performing arbitration and then selecting, from the lane line information from the plurality of sources, a first lane line corresponding to the reference lane line, where the first lane line and the reference lane line together form left and right lane lines of a present lane. The disclosure further relates to a road model generation device, a computer storage medium, an autonomous driving system, and a vehicle.