Lane Detection via Object Cluster Diagnostics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Driver assistance systems face challenges in identifying lane positions when road markings are obscured, as they rely on visual data from cameras, which may be unreliable in conditions like heavy traffic, snow, or when markings are obscured by other vehicles or objects.
Innovation Solution
The system clusters object tracks to optimize the identification of lane positions by calculating diagnostics for each object cluster, identifying rogue clusters, and reassigned measurements to improve the accuracy of lane detection, even in conditions where visual data is insufficient.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the system relies on visual data from cameras to identify lane positions, then the system can operate with simple sensor equipment, but the reliability of lane detection deteriorates when road markings are obscured by heavy traffic, snow, or other vehicles
Solution Approach 1:
The patent uses detected objects (vehicles, pedestrians) as intermediary elements to infer lane positions. Instead of directly detecting obscured road markings, the system clusters object positions and trajectories to indirectly determine lane boundaries and drivable paths, bypassing the harmful effect of obscured markings
Solution Approach 2:
The patent replaces the optical detection mechanism (camera visual detection of road markings) with a computational clustering mechanism that processes object position data to infer lane structure, substituting direct visual detection with indirect inference through object behavior patterns
2Reliability
If the system uses object clustering to identify lane positions, then the reliability of lane detection improves in obscured conditions, but the device complexity increases due to additional processing requirements
Solution Approach 1:
The system uses the objects themselves (vehicles, pedestrians) to define the lane structure through their natural clustering behavior. The objects' positions and trajectories self-organize into clusters that reveal lane positions, eliminating the need for complex external lane detection algorithms
Solution Approach 2:
The object clustering mechanism serves multiple functions: it identifies lane positions, determines drivable paths, and adapts to various road conditions simultaneously. This multi-functionality reduces the need for separate specialized systems for different detection scenarios
3Measurement precision
If the system removes rogue object clusters to optimize clustering, then the measurement precision of lane positions improves, but the loss of information increases due to removal of potentially valid data
Solution Approach 1:
The system applies partial removal of data by selectively eliminating only the excessive portions of rogue clusters (outlier measurements) while preserving the core valid data within those clusters. This partial action maintains measurement precision without unnecessarily discarding useful information
Data Source
AI summary
An apparatus for a motor vehicle driver assistance system is provided. The apparatus is configured to optimise object clusters, where each object cluster includes a sequence of position measurements for at least one object in the vicinity of the vehicle. Initially, in a pre-clustering phase, the assignment of the measured object positions to the object clusters may be based on the relative proximity of the measured object positions. The apparatus identifies a rogue object cluster on the basis of a first diagnostic, and a rogue object track from the measurements within the rogue object cluster. The position measurements from the rogue object track are removed from the clusters, and remaining position measurements in the rogue object cluster are reassigned to the other object clusters. The rogue object cluster is removed. Thus the object clusters are optimised.