Bringing intelligence into products for adaptive, dependable, and sustainable performance
The vision of Area Cognitive Products is to bring cognitive functions into industrial applications and products to enhance them, increase their usefulness and functionality, make them more robust and dependable, and make them more energy-efficient, adaptable, and sustainable. The extended vision is that cognitive products not only enhance and imitate the cognitive capabilities of humans to assist us in our daily work and life but also become cognitive entities themselves that cooperate with humans.
Research Approach
Our research is directly motivated by real-world industrial problems and challenges. We engage in a deep technical discussion with our industrial partners and scientific experts to gain valuable insights. These insights are crucial for our empirical experiments on testbeds to measure key performance indicators and produce reliable results. Additionally, with simulations and prototypes we demonstrate the efficiency of our achievements. The benchmarks and comparative analyses are conducted to optimize our approaches. Trade-off analyses ensure our solutions are robust, scalable, energy-efficient, and dependable under diverse operational conditions, enhancing performance and sustainability.
Technologies and Innovations
- Human-Centric Cognitive Functions: cognitive capabilities that prioritize human safety, quality of work, and ethical considerations.
- Adaptive Low-Energy Operation: energy-efficient and adaptive systems that intelligently scale power consumption.
- Flexible and Adaptive Composition: products that are compositional and reconfigurable.
- Enhanced Embedded Devices: advanced diagnostics, maintenance, localisation, and communication algorithms.
- Robust and Resource-Efficient AI for Edge Devices: machine learning models optimized for constrained hardware.
- Formal Verification for Software Correctness: Provide provably correct code for cognitive embedded software, ensuring reliability, safety, and sustainability.
Industries
We focus on transforming various industrial sectors through cognitive solutions:
- Automation Industry: Developing advanced safety concepts and flexible manufacturing solutions for highly automated production systems.
- Smart Manufacturing / Industrial IoT: Integrating cognitive functions into manufacturing processes and products, leveraging IoT (Internet of Things) and Industry 4.0 principles.
- Automotive Embedded Industry: Enhancing vehicle safety, functionality, and in-vehicle communication, including areas like wireless communication and dependability.
Topics
This research explores innovative methods for producing and collecting data from industrial processes, including both newly built and retrofitted systems. By applying AI and ML technologies to this data, we aim to enable more informed conclusions and decision-making that enhance safety, reliability, and sustainability across industrial ecosystems. To ensure the trustworthiness of these technologies, we address challenges such as bias and adversarial robustness. Furthermore, by leveraging transfer learning, few-shot, and continuous learning, we enable adaptive, on-device intelligence—allowing systems to learn from limited data, respond to changing conditions, and operate efficiently with reduced computational and energy demands.
This area concentrates on developing robust, efficient, and adaptable embedded systems specifically for cognitive and sustainable edge applications. We are bringing intelligence to the edge devices through the deployment of AI models. We tackle the challenges of constrained hardware by investigating techniques to make complex functionalities runnable on tiny devices, balancing system size, operational accuracy, and overall robustness.
This theme supports the entire embedded products development lifecycle by providing tools and frameworks that streamline development, deployment and reduce post-deployment issues for engineers. To this end, our work focuses on providing correctness guarantees for cognitive embedded software updates and automating its portability to new hardware architectures. Our methodology leverages formal methods to generate provably correct code, enhancing software maintainability and hardware reusability to boost the dependability and sustainability of future systems.
This theme is dedicated to enhancing the awareness and dependability of embedded devices for fields such as autonomous systems, automotive systems, and dynamic wired and wireless communication. Our research encompasses localisation and communication algorithms, addressing RF-coexistence issues, and developing methods that ensure security and safety guarantees under diverse conditions. We focus on low-energy and highly-adaptive wireless technologies like Ultra-Wideband (UWB) and Bluetooth Low Energy (BLE), enabling energy-efficient operation through low-power modes and energy harvesting.
Projects
