Very Efficient Deep Learning in IoT project with RISC-V and Renode


Topics: Open hardware, Edge AI

We are happy to announce our involvement in ‘Very Efficient Deep Learning in IoT’ (VEDLIoT) - a project funded by the European Commission and coordinated by Bielefeld University’s CoR-Lab, launched at the end of 2020. Comprising a 12-member international research group, VEDLIoT aims to develop a next-generation software/hardware platform for the Internet of Things. Antmicro’s contributions, among other things, will be to leverage its leading position in RISC-V, machine learning and simulation, developing open source RISC-V-based soft SoC infrastructure for the project and providing a Renode simulation environment for testing the platform’s security and robustness.

The project, which is still open to contributors, has received an 8 million EUR grant from the EU’s Horizon 2020 funding programme for research and innovation and is scheduled to be completed by the end of 2023.

VEDLIoT logo

“Computer and IoT systems are getting more and more efficient. This is enabling us to solve more challenging problems and accelerate automation in order to improve our quality of life,” explains Professor Dr.-Ing. Ulrich Rückert, who is the coordinator of the new VEDLIoT project and heads the Cognitronics and Sensor Systems research group at Bielefeld University. “But the volume of the data that is collected and processed is enormous - and the computing power required for this is very high. In addition, the algorithms are often too complex to quickly generate solutions in an appropriate amount of time.” VEDLIoT’s goal is to meet those requirements with an innovative, scalable and multi-purpose IoT-oriented platform.

Teaching the Internet of Things to learn

Striving to find a way to effectively handle the increasing complexity and huge amounts of data, VEDLIoT proposes a new method using distributed AI and deep learning algorithms. The use of deep learning enables the VEDLIoT platform to learn autonomously, which translates into increased performance and higher energy efficiency of IoT devices.

VEDLIoT smart mirror
Researchers on the new VEDLIoT project are developing a modular hardware platform that could be used in a range of applications, from an intelligent mirror to Smarthome devices. Photo: Bielefeld University/S. Jonek

Smart IoT applications are permeating more and more aspects of daily life as well as revolutionize various industries, performing advanced tasks such as tracking and object detection, navigation, industrial automation, inspection, sorting etc. They often involve neural networks that are trained to make specific decisions, while their design must ensure flawless operation in various, often harsh, in-field conditions. Antmicro builds this type of edge AI systems using a range of open source tools, software and hardware, and creates own data sets used for training neural networks to execute complex tasks. The portability and scalability of open source solutions as well as modern, software-driven development methodologies help us deliver better and more robust products and we are happy to carry those benefits over to the VEDLIoT project.

Building robust EdgeML applications with RISC-V and Renode

As part of the VEDLIoT project we are going to work on developing a portable and open source RISC-V based soft SoC infrastructure for system control and AI acceleration in FPGA. Relying on our expertise in this domain, we will be further developing the standard for custom AI acceleration in FPGA for RISC-V we have been involved in, as well as performing a survey of embedded machine learning frameworks to inform the selection of the tools to be used in the project.

The other task we will be in charge of is the creation of a Renode-based simulation platform that will enable development and continuous testing of the platform’s security features and its robustness. Our open source simulator offers a virtual and deterministic environment for software/hardware co-development, metrics for measuring the efficiency of workloads including ML, as well as CI-driven testing capabilities, and will be available to all project members and future users of VEDLIoT.

“As a RISC-V International Strategic Founding Member and an active developer of this open source ISA and its ecosystem, we are glad to be able to contribute to the project with our experience in using FPGA-based RISC-V solutions and simulation.” says Michael Gielda, Antmicro’s VP of Business Development. “VEDLIoT’s objectives align well with those of Antmicro: to enhance IoT and edge AI systems development with open source tools, workflows and platforms”. The company’s flagship open source simulator Renode has broad RISC-V support and will feature a simulated model of the FPGA SoC platform developed by Antmicro as part of VEDLIoT. “Both RISC-V and Renode have been enabling us to streamline and accelerate system development in various customer projects and collaborations, such as those with Google, Microchip, QuickLogic as well as other silicon and IP vendors, and we are excited to be bringing the benefits of both technologies to the project.”

Renode and RISC-V logo

Open source, scalable machine learning-oriented tools and IP are also a focus area of CHIPS Alliance that Antmicro is a Platinum member of. The RISC-V-oriented developments as well as enhancements in related tooling resulting from our participation in VEDLIoT will be an excellent addition to the ever-expanding collaborative ecosystem of CHIPS.

Our VEDLIoT-related efforts build on some of our earlier work around RISC-V support in Renode, open source SoC generators and open source tooling that we’ve been actively developing. As some of ourrecent projects, we have been building open source FPGA tools and Renode support for the exciting Core-V MCU, as well as enabling open source synthesis and simulation of complex SystemVerilog based designs.

Open call for additional project partners

Among project partners there are 7 universities and research institutes working in the area of artificial intelligence and the Internet of Things as well as companies of various sizes, ranging from our long-term partner, Research Institute of Sweden RISE to the automotive technology maker Veoneer or multinational corporation Siemens.

The project is still open to additional contributors, with at least ten more use cases expected to be financed as part of the programme. “Flowing into the IoT platform through the project, the new applications will allow us to make continuous improvements to the platform” says Dr. Carola Haumann, who is the project manager and Vice Managing Director of the CoR Lab. A prototype of the platform should be up and running by mid-2022.

If you are interested in learning more about using RISC-V and Renode to develop your next machine learning capable product, reach out to us at

See Also: