Embedded AI: Integration of AI algorithms in systems
Integrating artificial intelligence (AI) into control systems can both boost efficiency and significantly improve product functionality. In embedded systems, in particular, AI offers substantial added value, for example by analyzing incoming data, recognizing the status quo and making informed decisions based on this. Engineers developing innovative solutions also benefit from AI-driven fault detection, predictive maintenance and energy-efficient optimization.
To fully exploit the potential of self-learning AI algorithms, challenges such as processing large datasets and managing the associated high computing power must be addressed, all while taking into account all boundaries conditions such as accuracy, robustness and utilization of resources. This is exactly where we come in and offer our clients customized solutions to successfully integrate AI into products, while ensuring a reliable system.
A competent partner for embedded system design and AI development
With in-depth knowledge and experience in embedded systems, machine learning and both edge and cloud technologies, we combine proven approaches with cutting-edge methods. Our solutions range from the early conceptual stage to the development of an efficient AI architecture. Each step of the process is customized to meet our clients’ requirements, adhering to legal and industry standards such as the EU AI Act, MISR and ISO 8800.
Our services at a glance:
Concept development
Hardware and AI model architecture selection
Consultation and implementation of data management and infrastructure (MLOps, IS/IEC 42001)
Development and optimization of AI-based algorithms
Customized deployment and quality assurance solutions (deployment, safety, testing)
Our methodological expertise for your embedded AI solution
Resource-aware performance optimization
Embedded hardware often operates under limited computing resources, making the optimization of RAM, ROM, CPU time and energy consumption frequently the primary objective in AI development. The selection and pre-processing of data and functions helps to reduce complexity. Data processing is a strategic starting point to enable generalization on the one hand and to prevent overfitting on the other. Advanced training optimizations, such as well-chosen model architectures, quantization, pruning and the implementation of efficient inference pipelines reduce demand for computing resources.
Prototyping and validation
Early testing of embedded AI functionalities using prototypes under real-world conditions is essential to ensure reliability, performance and user satisfaction. This approach enables the timely identification of problems and improved resource utilization, leading to more effective and robust products. For real-world testing, algorithms are applied to the hardware in a non-intrusive manner without impacting the target system (shadow mode).
More intelligent control with AI
Parameter estimation: Tools like ITK SmartSearch can determine optimum system parameters, enhancing performance.
Supervised learning can improve the accuracy of state estimators or replace rule-based controls with AI systems that deliver comparable performance with lower resource requirements.
Reinforcement learning enables the development of high-performance control systems from scratch, requiring less domain knowledge.
Hybrid approaches that combine AI with traditional methods like Model Predictive Control effectively combat non-linearity and uncertainty.
A look at our reference projects
Reinforcement learning for 2-Wheeler
The challenge: Our client wants to enhance their current system, while accelerating development, by implementing a resource-efficient, modern AI algorithm. The use of AI in the two-wheel vehicle industry presents numerous safety-critical requirements and very limited computing resources on control units, compounded by frequent system changes.
Our solution: To develop non-linear driving functionality, we utilized a closed-loop simulation to train an AI-based control algorithm using reinforcement learning. Setting up an edge device connected to the control unit facilitated rapid prototype testing including real-world retraining to minimize simulation deviations. Deployment and hardware-specific optimization of the trained AI were achieved using the ETAS Embedded AI Coder to ensure compliance with safety standards such as ISO 26262 and MISRA.
Added value for the customer: This easily adaptable simulation and real-world setup, including the ETAS Embedded AI Coder, enabled our client to train, optimize and deploy an AI algorithm while reducing software complexity.
From research to application: Production-ready AI models for embedded AI systems
The challenge: The client needs to create a balanced dataset from predominantly homogeneous data for AI training and evaluation. Given the AI’s broad application across various product configurations and environmental conditions, good model generalization is crucial. The model must run on an embedded system with strict hardware constraints, and a comprehensive picture of performance and reliability under real-world conditions is necessary.
Our solution: A customized dashboard displays statistics of the existing data and uses an entropy algorithm to create a balanced dataset for training. Targeted analyses and experiments distil data features relevant for generalization. Based on this, new, adapted data campaigns are planned and implemented. The AI model is specifically optimized for the target hardware and deployed. To minimize testing efforts, shadow mode is employed, where the model is integrated into the end customer’s product and fed real sensor data without influencing AI outputs. Potential anomalies are then uploaded to the cloud, enabling automated AI improvements.
Added value for the customer: An AI that runs robustly for all use cases, leading to cost savings and reliable operation. Collecting relevant data saves unnecessary, expensive data campaigns and generates real added value. Extensive real-world testing fosters confidence in the AI application.
Benefits
Expertise in Embedded, Edge and Cloud Technologies
AI Development and Validation
Many years of cross-industry experience
Unsolved challenges? We look forward to your inquiry.
Expertise – Data Engineering & Artificial Intelligence
