Summary
- Profile Type
- Technology offer
- POD Reference
- TOIT20240312020
- Term of Validity
- 12 March 2024 - 12 March 2025
- Company's Country
- Italy
- Type of partnership
- Investment agreementResearch and development cooperation agreementCommercial agreement with technical assistance
- Targeted Countries
- All countries
Contact the EEN partner nearest to you for more information.
Find my local partner
General information
- Short Summary
- The Italian research team created a complete tool flow for the development of lightweight AI algorithms executable at the edge and custom power-efficient processing architectures for their execution. The IPs have been tested on several tasks including biosignal analyses such as EEG, EMG, and ECG. The researchers plan to continue the development of the project and look for investors and partners to collaborate on follow-up projects, both related to technology development and application research.
- Full Description
-
At-the-edge Artificial Intelligence (AI) empowers Machine Learning (ML) and Deep Learning (DL) at the network's periphery, closer to sensors and actuators, for localized data collection and processing, reducing latency, enhancing data privacy and security, and diminishing the need for cloud connectivity.
The Academic research group investigates such challenges, to develop smarter processing solutions at the edge of creating fresh electronic parts and systems, refining processing setups, improving connectivity, and developing software, algorithms, and middle-layer technologies
For example, recent research results have studied the usage of FPGAs at the micro-edge, for the inference of power-efficient event-based algorithms such as Spiking Neural Networks (SNNs), combining light-weight topologies with low-power
reconfigurable devices, to enable inference in an envelope of a few milliwatts. b) they have used tiny Transformer architectures for the near-sensor analysis of data streams, optimizing the algorithm to enable its execution on parallel microcontrollers.
They also created the prototype of a wearable sensor for ECG signals to monitor and identify real-time arrhythmias: The use of artificial neural networks will allow the maintenance of a high level of accuracy in the classification of arrhythmic pathologies even in the presence of artifacts and non-ideal factors typical of low-cost implementations.
The group offers their expertise for commercial exploitation and follow-up projects focusing on:
• digital industry,
• energy,
• agri-food and beverage,
• mobility, and
• digital society - Advantages and Innovations
-
The team was able to develop an ultra-low-power and low-cost wearable sensor for biosignals tested on different public datasets.
The solution is based on the latest generation neural networks such as transformers and spiking neural networks, runs on low-processing systems consumption, and allows to obtain accuracies aligned with state of the art in an envelope of few milliwatts - Stage of Development
- Available for demonstration
- Sustainable Development Goals
- Goal 3: Good Health and Well-beingGoal 7: Affordable and Clean Energy
Partner Sought
- Expected Role of a Partner
- Research centers, universities, and partners to collaborate in the engineering of the hardware and further exploitation and follow-up projects
- Type and Size of Partner
- Other
- Type of partnership
- Investment agreementResearch and development cooperation agreementCommercial agreement with technical assistance
Call details
- Coordinator required
-
Yes
Dissemination
- Technology keywords
- 01003003 - Artificial Intelligence (AI)01003024 - Cloud Technologies06001013 - Medical Technology / Biomedical Engineering
- Market keywords
- 03004003 - Other electronics related equipment03004002 - Components testing equipment03003 - Power Supplies05007007 - Other medical/health related (not elsewhere classified)
- Targeted countries
- All countries