AI-Generated Test Data for Image-Based Control Algorithm Coverage

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Complex image-processing algorithms, such as those in advanced driver assistance systems and autonomous vehicles, are difficult to test thoroughly due to the complexity of designing quality tests and achieving required test coverage, often requiring human intervention and manual creation of test cases.

Innovation Solution

The method employs artificial intelligence, specifically generative adversarial networks (GANs) and LSTM networks, to generate realistic synthetic image and sensor data, allowing for automated and extensive testing of algorithms by conditioning the AI on real data sets, thereby reducing human interaction and increasing test coverage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual test case creation with human intervention is used, then test quality can be maintained, but testing complexity and time consumption increase significantly

Engineering Contradiction:
Improvetest qualityVSAvoidtesting time consumption
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent uses Generative Adversarial Networks (GANs) to generate synthetic test data that copies the statistical properties and patterns of real-world data. The GAN creates artificial image sequences and sensor data that mimic real driving scenarios, allowing automated testing without manual intervention while maintaining test quality through realistic data distribution.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent transforms the testing process by changing parameters from manual creation to automated generation. By using trained AI models to generate test data with controlled parameters (image sequences, sensor values, action corridors), the system achieves both high test quality and reduced time consumption through systematic automated generation.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If comprehensive testing coverage is pursued for complex algorithms, then algorithm reliability improves, but test design complexity increases

Engineering Contradiction:
Improvealgorithm reliabilityVSAvoidtest design complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements self-service testing where the system automatically generates its own test data and performs validation without external intervention. The GAN-based framework autonomously creates diverse test scenarios, executes algorithms, and evaluates results, achieving comprehensive coverage while simplifying test design through automation.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent creates a universal testing framework that handles multiple algorithm types (image processing, sensor processing, control algorithms) through a single GAN-based system. The framework generates diverse test data including image sequences, sensor data, and action corridors, making it applicable to various autonomous vehicle algorithms without requiring separate test designs.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Reliability

If real-world test data collection is used, then test realism is maintained, but data diversity and test coverage are limited

Engineering Contradiction:
Improvetest realismVSAvoidtest data diversity
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent introduces dynamics by using GANs to generate continuously varying test scenarios. The system can dynamically create diverse driving conditions, weather patterns, and environmental scenarios that go beyond collected real-world data, maintaining realism through learned data distributions while achieving unlimited diversity through automated generation.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3782081B1Method for generating a test data set, method for testing, method for operating a system, device, control system, computer program product, computer readable medium, generation and use
Publication Date: 2022.12.07 SIEMENS IND SOFTWARE NV
  • EP3782081B1 patent drawingFigure 1
  • EP3782081B1 patent drawingFigure 2
  • EP3782081B1 patent drawingFigure 3

AI summary

The invention relates to a method for producing a test data record (12) for an algorithm (1), comprising the following steps: 1.1 a trained artificial intelligence (8) is provided, 1.2 the artificial intelligence (8) is stimulated, particularly using a random signal (11) and/or using a quasi linearity property for visual concepts, and 1.3 the artificial intelligence (8) produces at least one test data record (14), which comprises image data (15) and action regions (16) associated with the image data (1g) and/or sensor data and action regions associated with the sensor data. The invention furthermore relates to a method for operating a system for the automated, image-dependent control of a device and an apparatus for carrying out the aforementioned method. Finally, the invention relates to a control system for a device which comprises such an apparatus, and a computer program product, a computer-readable medium, the production of a data storage device and the use of an artificial intelligence (8).