Lane Change Indication Using Position Windows and CNN Classification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for determining lane changes in autonomous vehicles face challenges due to sensor inaccuracies, leading to oscillations in lane association and requiring extensive manual annotation of training data, which is time-consuming and costly.

Innovation Solution

A method using a lane change model and position data to determine lane changes, incorporating a convolutional neural network (CNN) for accurate classification, and reducing data processing complexity by utilizing a two-dimensional projection of the surroundings model, thereby minimizing human interaction and increasing accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If fixed rules are used to automatically annotate lane change data, then the annotation process becomes fully automated, but sensor inaccuracies cause oscillation in lane association leading to false lane change detections

Engineering Contradiction:

Improveautomation of lane change annotationVSAvoidaccuracy of lane change detection

Core Design Contradiction:

Extent of automationVSReliability

Solution Approach 1:

The system performs preliminary actions by defining a evaluation window before making a lane change determination. Multiple position evaluations are collected within this time window before a final lane change indication is generated, allowing the system to pre-assess positional stability before committing to a lane change detection

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements periodic evaluation by assessing vehicle position at multiple discrete time points within the evaluation window. This periodic sampling allows the system to observe positional patterns over time and distinguish true lane changes from sensor-induced oscillations

Inventive Principle:

Principle #19Periodic action

2Measurement precision

If manual annotation of training data is performed, then the accuracy of training data increases, but the time and cost required for data preparation increases significantly

Engineering Contradiction:

Improveaccuracy of training data annotationVSAvoidtime required for data annotation

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs self-service by using its own sensor data and the developed lane change model to automatically generate annotated training data. The model evaluates position data and generates lane change indications that serve as ground truth labels, eliminating the need for external manual annotation while maintaining high accuracy through the model's periodic evaluation mechanism

Inventive Principle:

Principle #25Self-service

3Reliability

If multiple sensor data are fused to improve surroundings model accuracy, then the reliability of object recognition increases, but the complexity of data processing and lane change determination increases

Engineering Contradiction:

Improveaccuracy of surroundings modelVSAvoidcomplexity of lane change determination system

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system extracts only the necessary position data from the comprehensive surroundings model for lane change determination. By focusing specifically on vehicle position relative to lane center and evaluating this extracted data through the lane change model with periodic assessments, the system avoids processing the full complexity of fused sensor data while maintaining accurate lane change detection

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS12012102B2Method for determining a lane change indication of a vehicle

Publication Date: 2024.06.18 BAYERISCHE MOTOREN WERKE AG

AI summary

A method for determining a lane change indication of a vehicle, preferably a passenger car, includes the following steps: loading a lane change model and position data of the vehicle, the position data indicating a distance of the vehicle from a lane center of a lane; and determining a lane change indication using the position data and the lane change model.