The EdgeAI project presented its latest findings on heterogeneous edge AI solutions at the European Nanoelectronics Applications, Design & Technology Conference (ADTC) and edaWorkshop24 held in Dresden, Germany (April 9-10, 2024). The session, titled “Heterogeneous Edge AI Solutions Across Industrial Sectors,” was part of Conference Session 4: Edge AI.

With the increasing demand for edge AI applications across industrial sectors with diverse requirements, specifications, and computational needs, there is an evident necessity for a revitalised computing architecture custom-designed for edge AI. This includes the integration of multiple heterogeneous mixes of processors as functional blocks like neural processing units (NPU), central processing units (CPU), and graphics processing units (GPU).

By using the appropriate processor for the different edge AI functions, heterogeneous computing maximises application performance, efficiency, and battery life to enable new and enhanced edge AI uses in industrial sectors like energy and agri-food.

The EdgeAI session at ADTC 2024 featured presentations on various aspects of edge AI:

  • Siracusa: a 16nm Extended Reality PULP SoC with Heterogeneous At-MRAM Computing – Francesco Conti (University of Bologna, IT)
  • Power-Aware Computing for the Internet of Things – Pavel Smrz (COGNITECHNA, CZ)
  • Edge-AI-Driven Real-Time Clustering for Threat Detection in 6G Networks – Abdelghani Bourenane (Scuola St.’Anna, IT)
  • Heterogenous Edge AI Solutions Across Industrial Sectors – Ovidiu Vermesan (SINTEF, NO)

Stay tuned for more!

Share This