Similar Scenario Detection Using Reduced Sensor Feature Embeddings

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for determining similar scenarios from sensor data in driver assistance systems and automated driving are computationally intensive and require significant resources, necessitating a more efficient approach to identify relevant scenarios.

Innovation Solution

A method utilizing machine learning algorithms with encoders and optimization techniques to generate dimensionally reduced feature representations from sensor data augmentations, clustering similar interaction maneuvers by minimizing similarity loss, and applying LSTM layers to process trajectory data effectively.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If existing methods for determining similar scenarios from sensor data are used, then comprehensive scenario analysis is achieved, but computational resources and processing time are excessively consumed

Engineering Contradiction:
Improvescenario identification efficiencyVSAvoidcomputational resource consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent extracts and processes only the most relevant features from sensor data using selective augmentation and feature extraction techniques. By taking out only the critical scenario-determining features rather than processing all raw sensor data, the system achieves comprehensive scenario analysis with reduced computational resource consumption.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent segments the scenario determination process into distinct stages: data augmentation, feature extraction, similarity calculation, and scenario classification. This segmentation allows each stage to be optimized independently, improving overall productivity while managing computational resources efficiently at each step.

Inventive Principle:
Principle #1Segmentation

2Reliability

If comprehensive sensor data processing is performed to identify all relevant scenarios, then scenario coverage is improved, but processing time increases significantly

Engineering Contradiction:
Improvescenario coverageVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent performs preliminary data augmentation and feature extraction before actual scenario matching. By preparing augmented datasets and extracting key features in advance, the system ensures comprehensive scenario coverage during execution while minimizing real-time processing time.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent applies selective data augmentation techniques that generate additional training samples only for underrepresented scenario types. This partial action approach ensures comprehensive scenario coverage for critical cases while avoiding unnecessary processing of already-well-represented scenarios, thus reducing overall processing time.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4191469B1Method for determining similar scenarios, training method and training control device
Publication Date: 2026.02.11 DSPACE SE & CO KG
  • EP4191469B1 patent drawingFigure 1
  • EP4191469B1 patent drawingFigure 2~3

AI summary

The invention relates to a computer-implemented method for providing a machine learning algorithm for determining similar scenarios based on scenario data from a dataset (D) of sensor data, wherein an optimization algorithm (A3) is applied to the feature representation (18) of the first augmentation (14) of the dataset (D) of sensor data output by the first machine learning algorithm (A1), and wherein the optimization algorithm (A3) approximates the feature representation (20) of the second augmentation (16) of the dataset (D) of sensor data output by the second machine learning algorithm (A2). The invention further relates to a method for determining similar scenarios based on scenario data from a dataset (D) of sensor data and a training controller.