Fraunhofer IKS Playground

The Fuzzy Cause Tree (FCT) allows you to analyze cause-and-effect relationships between triggering conditions and system failures with greater precision, especially in the face of uncertainties. The Fuzzy Cause Tree (FCT) is an innovative approach developed by Fraunhofer IKS that significantly reduces risks and improves the safety of the intended functionality (SOTIF).

What is the Fuzzy Cause Tree method?

The Fuzzy Cause Tree method uses the principles of fuzzy logic to manage the complexity of real-world systems. Unlike traditional logic, which is limited to absolute true or false values, fuzzy logic allows for a differentiated assessment of truth values on a spectrum. This makes it possible to capture uncertainties and vague conditions more effectively and make smarter decisions.

Your benefits with the Fuzzy Cause Tree:

• Leverage the power of fuzzy logic: Our approach captures uncertainties and vague conditions far better than conventional methods.
• Enhance your analysis with FCTs: Combine the FCT method with statistical methods such as Monte Carlo simulations to gain deeper insights into the effects of causes on system behavior. This enables you to make informed decisions that significantly improve safety.

Why the Fuzzy Cause Tree?

The Fuzzy Cause Tree offers a flexible and intuitive approach to modeling complex causal relationships that goes beyond the limitations of conventional methods. It enables a better understanding of risks and promotes informed decisions in safety-critical systems, especially in the automotive industry.

Try the fuzzy cause tree method now!

Robuscope, an application developed by Fraunhofer IKS, allows you to easily evaluate the reliability of your AI models. With just one click, you can gain valuable insights into the robustness and performance of your AI systems.

What is Robuscope?

Robuscope is a self-service platform that answers the question: "Can I trust my AI?" Robuscope provides you with detailed information on potential improvements to ensure that your AI models are reliable, efficient, and ready for use in safety-critical applications.

Your benefits with Robuscope:

• No sensitive data required: You don't need to disclose any confidential or sensitive information to test your systems for reliability.
• Actionable insights: Get clear interpretations and actionable results that help you optimize your AI model.
• Comprehensive assessment: Detailed reports on the robustness of your AI models reveal potential weaknesses and areas for improvement, ensuring that your AI systems reach their full potential.

Why Robuscope?

With Robuscope, the days of AI black boxes are over. Robuscope enables you to address potential issues early on and increase the trustworthiness of your AI applications. This allows you to minimize risks and maximize performance.

Test your AI models now!

The Fraunhofer IKS Safety Assistant is an innovative front-end solution that optimizes safety engineering tasks through the use of generative AI (GenAI). Given the increasing complexity of software development for safety-critical systems, AI-powered assistants can provide support in overcoming these challenges.

What is the Fraunhofer IKS Safety Assistant?

The Safety Assistant leverages the capabilities of large language models to bridge the gap between technological progress and safety assurance. To do this, it automates essential aspects of the safety engineering workflow, thereby changing the approach to safety engineering.

Your benefits with the Fraunhofer IKS Safety Assistant:

• Better understanding of the safety engineering workflow: Our mission is to deepen the understanding of complex relationships within safety-critical systems. The Safety Assistant examines how generative AI can streamline the planning of measure validation and ensure the accurate implementation of requirements.
• Generate robust safety verifications: With the help of the Safety Assistant, safety engineers can generate robust safety verifications and bring innovations into application quickly and safely.

Why choose the Fraunhofer IKS Safety Assistant?

The Fraunhofer IKS Safety Assistant enables you to master the complexity of software development for safety-critical systems. This allows you to bridge the gap between innovation and safety.

Learn about the Fraunhofer IKS Safety Assistant vision now!

With the Fraunhofer IKS FAST Framework, you can effectively and efficiently handle decision-making tasks driven by visual information such as quality inspections or medical diagnoses – even when only limited data is available. 

What is the FAST Framework?

FAST stands for "Feedback-guided Automation of Sub-tasks." The framework improves decision-making through the use of advanced automation techniques. To do this, FAST identifies sub-tasks that can be automated, while the remaining more complex tasks are delegated to experts.

Your benefits with the FAST framework:

•  Immediate impact: Rapid implementation of partial solutions without lengthy data collection phases.
• Lower barriers and risks: Minimal initial data requirements make setup easier.
• Enhanced efficiency: Experts can focus on complex problems.
• Build trust: Collaboration with the system promotes acceptance and trust in AI solutions.

Why FAST?

The FAST framework enables increased operational efficiency, reduced errors, and improved overall productivity by combining the strengths of automation with human expertise. This hybrid approach not only speeds up decision-making but also ensures high-quality results.

Learn more about FAST now!

With Fraunhofer IKS's AS-PHALT proof of concept, you can use the power of AI-driven large language models (LLMs) to perform hazard analysis and risk assessment (HARA). This enables you to use resources more efficiently and improve the reliability of safety assessments.

What is AS-PHALT?

AS-PHALT stands for “Automation Systems – Pro Human Assistant LLM Tool.” Using the example of hazard analysis and risk assessment, we illustrate how AI can help identify and evaluate hazards through an interactive, user-friendly interface. The concept aims to optimize the hazard analysis process through innovative automation techniques.

Your benefits with AS-PHALT:

• Immediate effect: Experience the rapid deployment of a partial solution that bypasses lengthy data collection phases.
• Fewer obstacles and risks: Benefit from minimal initial data requirements that facilitate integration into existing processes.
• Improved efficiency: Relieve safety experts by automating routine tasks so they can focus on complex problem solving.
• User-friendly interface: Work efficiently with the system thanks to a structured workflow and familiar visualization of results.

Why AS-PHALT?

The AS-PHALT concept makes it possible to increase efficiency in safety technology and reduce the complexity of safety assessments in automated applications. The use of generative AI makes safety assessment more agile and innovative.

LEARN MORE ABOUT AS-Phalt!

Introducing the Enhanced CADx Framework, an advanced computer-aided diagnosis (CADx) system developed to empower clinicians with AI support that is transparent, interpretable, and easy to validate. Moving beyond black-box diagnostics, this framework provides multi-layered explanations – both visual and concept-based – through a single intuitive interface.

What is CADx?

The Enhanced CADx Framework bridges the gap between clinical expertise and artificial intelligence by making AI reasoning clear, explainable, and clinically meaningful. Rather than providing opaque outputs, the system enables clinicians to explore how and why the AI arrived at its diagnosis –supporting trust, efficiency, and collaboration in healthcare workflows.

Your benefits with CADx:

• Multi-modal explainability: Combines visual explanations (e.g., Grad-CAM heatmaps) with concept-based reasoning (e.g., TCAV) for a holistic understanding of AI decisions.
• Intuitive interface: Integrates multiple explanation types in one user-friendly platform, allowing clinicians to easily interact with and validate AI results.
• Clinically grounded insights: Translates AI reasoning into clinical concepts, ensuring that explanations are relevant and actionable.
• Aligned diagnostics: Guarantees that diagnostic outputs correspond directly to the explanations provided, enhancing consistency and transparency.

Why CADx?

Traditional CADx systems often operate as “black boxes”, offering limited insight into their diagnostic rationale. The Enhanced CADx Framework eliminates this opacity, reducing the need for manual validation and strengthening clinician confidence in AI-assisted healthcare. By revealing the logic behind AI diagnoses, the system transforms AI from a mere analytical tool into a trusted, collaborative partner in patient care.

 Learn more about cAdx

Analyze and enhance safety in industrial workspaces efficiently and pragmatically. Gain reliable insights into the interaction between humans and machines and establish a solid foundation for safe, productive processes.

With Safety Net, we support organizations in deploying AI-based person detection in safety-critical industrial applications and systematically reducing operational risks.

What is the Safety Net approach?

The Safety Net approach leverages modern artificial intelligence methods to reliably detect people in industrial environments and assess safety-relevant situations at an early stage. Instead of relying solely on rigid protective measures, Safety Net enables a context-aware and situation-dependent evaluation of risks.

By taking real-world operating conditions into account, uncertainties can be captured more effectively, and safety functions can be applied precisely where they are needed.

Your benefits with Safety Net:

• Reliable AI-based person detection: The approach enables precise detection of people in the vicinity of autonomous systems and significantly contributes to improved workplace safety.
• Systematic safety assessment: Safety Net supports a structured analysis of potential hazards and provides transparency regarding safety-relevant system states.
• Practical integration: The approach can be deployed alongside existing safety measures and supports efficient operations without unnecessary restrictions.

Why Safety Net?

Safety Net offers a flexible and transparent approach to safeguarding human–machine interactions in industrial environments. By combining AI-driven perception with structured safety reasoning, risks can be better understood and well-informed decisions can be made.

Especially in dynamic and automated industrial settings, Safety Net helps organizations balance safety and productivity while enabling the responsible use of artificial intelligence.

Learn More about Safety Net

With our Safety Assurance approach, you can analyze and substantiate the safety of AI-based vehicle dynamics functions throughout the entire development process – even under uncertain conditions and in open operating environments. The approach was developed at Fraunhofer IKS and supports you in reducing risks and providing a sound safety argument for the use of intelligent systems.

What is the Safety Assurance approach for AI systems?

The Safety Assurance approach extends established development processes in a targeted manner to make them suitable for the use of artificial intelligence. Instead of addressing safety only at selected points, it is considered systematically across all phases – from analysis and design to verification and validation.
This makes it possible to structure and justify even complex interactions between AI, the system, and its environment in a transparent and comprehensible way.

Your benefits with Safety Assurance for AI:

• Leverage a structured development approach: Integrating safety aspects into all development phases creates transparency and enables well-founded safety arguments.
• Strengthen your evidence through systematic assurance: Clear argumentation structures and evidence-based justification provide a robust foundation for approval, assessment, and further development of AI-based functions.

Why Safety Assurance for AI-based vehicle dynamics?

The Safety Assurance approach provides a clear and comprehensible framework for assuring complex AI systems. It helps developers identify risks at an early stage, implement targeted safety measures, and convincingly argue system safety even in open environments. Especially in safety-critical domains such as vehicle dynamics, this approach builds trust and supports confident decision-making.

Learn More about Safe AI Sensors

Quantum-enhanced artificial intelligence opens up new possibilities for more precise and efficient diagnostics in medical imaging. Our approach combines classical AI methods with quantum technologies to systematically address existing limitations—particularly in scenarios with limited training data. This work is developed at the Fraunhofer Institute for Cognitive Systems IKS with the goal of enabling powerful and trustworthy AI solutions for medical applications.

What is quantum-enhanced AI for medical diagnostics?

Quantum-enhanced AI leverages hybrid quantum–classical algorithms that combine the strengths of both computing paradigms. By integrating quantum-specific computational operations into classical AI models, complex patterns in medical imaging data can be captured more effectively.

This approach opens up new perspectives for the use of AI in medical diagnostics—even under challenging conditions such as limited data availability.

Your benefits with quantum-enhanced AI:

• Leverage innovative hybrid algorithms:
  The combination of quantum and classical processing enables improved analysis of complex medical imaging data.
• Increase efficiency when working with limited data:
  Quantum-enhanced approaches provide new ways to develop robust and high-performing AI models even with small datasets.

Why quantum-enhanced AI in medicine?

The use of AI in medical imaging comes with high requirements for accuracy, reliability, and data quality, while large, well-annotated datasets are often difficult to obtain. Quantum-enhanced AI offers a forward-looking approach to systematically address these challenges and unlock new performance potential for medical diagnostic systems.

Learn More about Quantum-enhanced AI for Medical Diagnostics

The Quantum Zoo offers an intuitive and interactive entry point into the world of quantum computing. The approach makes it possible to understand fundamental concepts and mechanisms of quantum-based computation. The Quantum Zoo is developed at Fraunhofer IKS with the goal of making complex quantum technologies tangible and accessible. 

What is the Quantum Zoo?

The Quantum Zoo is a playful, interactive environment for demonstrating gate-based quantum computing. It illustrates how quantum operations act on quantum states that can represent more than one classical state at the same time.
By using suitable algorithms, it demonstrates how quantum parallelism can be leveraged to perform computations that can theoretically outperform classical computing methods exponentially.

Your benefits with the Quantum Zoo:

• Experience the principles of quantum computing first-hand:
  Core concepts such as superposition, interference, and entanglement are conveyed visually and interactively, making them intuitive and easy to understand.
• Deepen your understanding of quantum circuits:
  Users can build simple quantum circuits themselves, combine quantum gates, and directly observe how these operations affect the states of the qubits.

Why the Quantum Zoo?

Quantum computing holds enormous promise, but it also presents significant challenges—particularly when mapping real-world problems onto efficient quantum algorithms. The Quantum Zoo provides a low-threshold yet technically sound introduction to this field. It helps users gain a better understanding of how quantum-based systems work.

Learn More about Quantum-ZOO

The QuaST Decision Tree opens up new possibilities for efficiently implementing quantum-enhanced solutions to complex optimization problems. It introduces a structured abstraction layer that enables the automated selection and execution of suitable algorithms – without requiring deep expertise in quantum computing. The approach is designed to deliver practical, high-performance solutions across domains such as logistics, finance, and supply chain management.

What is the QuaST Decision Tree?

The QuaST Decision Tree is a modular framework for systematically developing quantum-enhanced optimization solutions. It organizes the entire workflow – from problem formulation to result interpretation – within a clear and structured decision logic.

During a forward pass through the decision tree, an appropriate quantum-based approach is automatically selected and executed on simulators or quantum hardware. In a subsequent backward pass, results are processed and translated into actionable, application-specific solutions.

Your benefits with the QuaST Decision Tree:

• Simplified access to quantum-enhanced solutions:
  The structured framework reduces complexity and allows users to focus on their optimization challenges rather than technical implementation details.
• More efficient development and deployment:
  Automated decision-making minimizes iterative trial-and-error processes, saving both time and resources.
• High flexibility and seamless integration:
  The framework can be easily adapted to specific use cases and integrated into existing workflows and external tools.

Why automated quantum-enhanced optimization?

Developing quantum-enhanced solutions typically requires interdisciplinary expertise, ranging from algorithm design to implementation and tuning.
At the same time, organizations need efficient and scalable ways to leverage these emerging technologies.

The QuaST Decision Tree addresses this challenge by bridging the gap between application and technology – enabling a structured, automated pathway to unlock the full potential of quantum computing.

Learn More about QuaST Decision Tree

LLM-based troubleshooting assistance enables faster and more reliable root cause analysis in manufacturing environments. Our approach integrates relevant knowledge sources such as machine documentation, operational data, and domain expertise into an intelligent assistant that guides operators through complex diagnostic processes. This work is developed at the Fraunhofer Institute for Cognitive Systems IKS with the goal of making trustworthy AI-based troubleshooting accessible across industrial applications.

What is LLM-based troubleshooting assistance?

LLM-based troubleshooting assistance uses large language models combined with retrieval pipelines and knowledge graph integrations to provide operators with clear, context-aware guidance during production disruptions. By connecting live and historical data, machine logs, and codified domain knowledge, the assistant delivers tailored, real-time insights through a natural language interface, allowing operators to focus on solving the problem at hand rather than searching through documentation.

Your benefits with LLM-based troubleshooting assistance:

• Accelerate root cause analysis: Relevant knowledge from machine documentation, logs, and domain expertise is made instantly accessible, reducing time spent on manual investigation.
• Retain and distribute expert knowledge: Best practices and tacit knowledge are captured and made available to all operators, independent of individual experience levels.

Why LLM-based troubleshooting in manufacturing?

Production disruptions demand rapid response. Continuously changing manufacturing systems, complex processes, and varying levels of digitalization make it difficult to identify root causes quickly. LLM-based troubleshooting assistance addresses these challenges by combining dependable AI guidance with real-time data integration, enabling more flexible, downtime-reduced production.

Learn More about Troubleshooting dashboard