Edge AIoT (Artificial Intelligence integrated in the Internet of Things) is not only a trend, but also increasingly necessary for our industry. More often, small devices need to process huge amounts of data close to the point of origin with low latency, while efficiently running AI algorithms. For this, solutions that process the data in the sensors themselves are already on the rise, but these require a comprehensive and innovative approach to minimize costs and power consumption. Therefore, in the context of industrial applications, continuous improvement of hardware for computing AI algorithms is needed. This challenge is addressed by 19 companies and universities in the context of the research project CLEVER[2]. Their common goal is the development of an energy-efficient hardware accelerator for AI algorithms.

The contribution of the Fraunhofer IMS: Development of an energy-efficient hardware compiler

As part of the project, the partners work on various innovations to realize an efficient operation of hardware for processing AI. The Fraunhofer IMS focuses on the development of an EDA tool (Electronic Design Automation), which should simplify the design of an accelerator. For this task, a hardware compiler is to be designed as a component for the tool. This is a software that translates a given AI algorithm directly into HDL code (Hardware Description Language) for the accelerator.

Thereby, the Fraunhofer IMS team focuses on the subarea of neural networks. In the next step, the completed accelerator is to be connected to the AIRISC, a RISC-V processor developed by Fraunhofer IMS, which provides the necessary openness for integration. In doing so, the local AI shall reduce the amount of data that needs to be sent. At the same time, the resulting accelerator would allow AI algorithms to run more energy-efficiently on the smallest embedded devices.

The efficient hardware can enable numerous application scenarios

The benefits of optimized hardware would be reflected in a wide range of applications, where the consortium will focus on the following use cases in the CLEVER project:

1) Digital twins for in-factory optimization.

2) Augmented Reality (AR) for shopping malls.

3) Smart agriculture for high-yield eco-farms.

For example, in the field of agriculture, battery-powered and camera-based systems could analyze images on site and send an alarm if necessary to precisely adjust the cultivation of fields with the resources used, such as water or pesticides. The accelerator designed in the research project would contribute to more efficient analysis of data, resulting in longer battery life. By saving resources and optimizing cultivation in general, it would promote higher-yielding as well as more sustainable agriculture.

Cooperating institutions

The CLEVER research project involves a consortium consisting of the following institutions: Sant'Anna School of Advanced Studies, Dell Technologies, Synopsys, Eccenca, Eindhoven University of Technology, Agricolus, Aalborg University, Gottfried Wilhelm-Leibniz University of Hanover, Cnit, BMW, Nvidia, Cortus, Golfe Engineering, Dresden University of Applied Sciences and Arts, National Technical University of Athens, Italtel, University College Cork, Innatera, Fraunhofer IMS.

Link to project website: https://www.cleverproject.eu/

 

Fraunhofer IMS

For over 30 years scientists at Fraunhofer IMS in Duisburg have been dealing with the development of microelectronic circuits, electronic systems, microsystems and sensors. Because of its comprehensive know-how, the access to technology and the high-quality development work, the Institute is a globally recognized partner for the industry. In four business units and core competencies each, Fraunhofer IMS is dedicated to applied research, advance development for products and their applications. High-quality, efficient and marketable technologies and procedures that are used in a very wide range of branches take center stage in contract work.

www.ims.fraunhofer.de/en.html

 

Research Fab Microelectronics Germany (FMD)

The Fraunhofer IMS is part of the Forschungsfabrik Mikroelektronik Deutschland (FMD) - a cooperation of the Fraunhofer Group for Microelectronics with the Leibniz Institutes FBH and IHP. As a pioneer for cross-site and cross-technology collaboration, the FMD addresses current and future challenges in electronics research and provides important impulses for the development of elementary innovations for the world of tomorrow.

https://www.forschungsfabrik-mikroelektronik.de/en.html | Also visit our virtual 3D showroom at https://fmd-insight.de/showroom

 

[1] KDT JU: Key Digital Technologies Joint Undertaking

[2] CLEVER is an acronym for: Collaborative edge cLoud continuum and Embedded AI for a Visionary industry of thE futuRe. The project is funded by the European Union under 101097560 - CLEVER - HORIZON-KDT-JU-2021-2-RIA and co-financed by the