Machine learning Embedded hardware
Ongoing

EmbedML | Accelerating the integration of machine learning in products with embedded hardware

Financed by

The European EmbedML project aims to accelerate the integration of machine learning in products with embedded hardware.

Context

Machine learning (ML) is on the rise as a key enabling technology to create and capture more value with smart products. While ML models are well known to be implemented on high-end platforms in the cloud, recent advancements in this domain allow to deploy ML models on tiny and ubiquitous microcontrollers. This embedded ML technology is disruptive for the way products can process data locally, find novel solutions to complex problems and achieve new levels of embedded intelligence. Combining ML with low-cost and low-power microcontrollers without relying on centralized data transfer is beneficial to a wide range of applications, from predictive maintenance in manufacturing to patient monitoring in healthcare. As embedded ML technology is now coming within reach of early adopters, their central challenge is how to integrate embedded ML in the development of their new products in the right way and at the right cost.

 

EmbedML integration machine learning in products with embedded hardware

Goal, approach and results

The main goal of the EmbedML project is to support SMEs with their product smartification by means of embedded ML, to clarify the problem-solution fit of their envisaged solution, accelerate their proof-of-concept (PoC) realization and facilitate the integration of embedded ML in their entire product development.

Therefore, EmbedML develops SME-oriented supporting tools to help the target group accelerate the integration of machine learning in their products with embedded hardware:

Online embedded ML case-book of best practice examples of embedded ML applications, that assists SMEs in the identification and evaluation of opportunities in embedded ML. 
Toolset for embedded ML PoC development supporting SMEs with their in-house realization of PoCs thanks to system design rules and best practices on available development platforms. 
Embedded ML development guidelines covering the different stages, engineering methods and required effort and expertise from initial feasibility to product follow-up in the field. 
Integrated development approach for Embedded ML, covering the entire development cycle. 
Industry-driven reference cases validating and demonstrating the integrated development approach. These cases are elaborated progressively from initial feasibility up to full system concept and PoC.

EmbedML will effectively transfer these project results to 160 companies by the end of the project through broad dissemination actions activating the entire target group and deep-dive seminars and workshops for in-depth knowledge transfer on embedded ML. Within two years after the project and with the support of EmbedML, 40 companies will have engaged in implementations, from feasibility studies and PoCs to development projects. 
Some relevant products and tools along the Embedded Machine Learning life-cycle

 

EmbedML integration machine learning in products with embedded hardware
Some relevant products and tools along the Embedded Machine Learning life-cycle

Target group

EmbedML targets both SME product companies and service providers active in the manufacturing industry who are developing sensors, monitoring solutions or smart systems, and having product smartification as a business driver. The project results will be developed in close interaction with a user committee of product companies and service providers.
Thanks to the EmbedML supporting tools product companies add more value to their products with embedded ML, reduce development costs and risks and speed up their product research. Service providers extend their service portfolio and benefit from market-creation. Involving both product companies and service providers fosters collaboration across the value chain. The largest impact is to be expected on the efficiency of vertical applications and on the automation of low-level monitoring tasks, shifting human work to higher-level decision making. 
Does your company wish to be part of the user committee? Let us know! Participation to the user committee is free, and in return for a limited investment of your time, you will gain access to the project results, and you can help steer the project and exchange experiences with the other participating companies.

Funding

  • Funding agency: VLAIO 
  • Reference & Funding framework: COOCK CORNET HBC.2021.0894

 

Partners

In collaboration with

Timing

Jul 2022 - Jun 2024

Our experts

Do you have a question?

Send them to innovation@sirris.be