Author: Marko Sagadin, Embedded Machine Learning Engineer at IRNAS

Image for post
Image for post

Introduction

I recently wrote a Master’s thesis on detecting elephants from thermal cameras, with the help of Machine Learning. Thesis is publicly available on my GitHub repository.

Classifying images has come a long way in the last decade; at this point, it is quite a norm in the field of web, mobile and desktop applications, however, it is not yet widely used in embedded systems.
Recent developments in the field of embedded Machine Learning, also known as TinyML, have pushed boundaries of what is possible. …


Power consumption optimization is at the heart of IRNAS’ development of advanced applied IoT solutions. As such, we are always working actively to optimize our workflows and automate testing with continuous integration. Following our guiding principles and the active development of solutions, based on Nordic Semiconductor products, we were very excited to learn about the newly introduced Power Profiler Kit II (PPK2).

Image for post
Image for post

This new generation of a very useful tool enables our developers to have a power consumption analysis tool handy at all times. While the graphical interface through nRF Connect application is very nice, we also like to use equipment in fully automated workflows. …


For the past two months, IRNAS and Smart Parks, in partnerships with Hackster.io, have been working on building the next generation OpenCollar ElephantEdge tracker, closely looking at requirements and features that make this a success. Two key concepts are driving it, rock-solid field performance and an intuitive user experience. In the previous blog and webinar, we did a deep-dive into the technical choices made in the design process and the experiences shared by Smart Parks, we hereby look deeper into the firmware and workflow aspects.

Together, we make sure to deliver sustainable, future-proof solutions. This requires a skillful integration of the latest technologies with years of development experience and a very agile hands-on process with the users, such that we converge to the best solution in the shortest amount of time. …


This decade started with a rapid growth of connectivity options such as BLE5, LoRaWAN, NB-IoT, satellite technologies and many other exciting options. In particular, the power consumption of the devices has decreased an order of magnitude on the processing and communication front, leading to the capability of acquiring a large amount of sensor data at very low power while being power and cost limited to only send a handful of measurements.

At IRNAS, we work with a global range of customers from industrial sensing in critical infrastructure to remote monitoring of animals all over the globe. …


IRNAS and Smart Parks have been working on designing the next generation of open-source tracking solutions for national park management and wildlife protection for the past two years and the deployment of these solutions in the field has proven very valuable.

Webinar

Image for post
Image for post

Event info

  • First OpenCollar for Elephant successfully deployed in Liwonde National Park — Park Rangers in Liwonde National Park in Malawi are for the first time able to see elephant locations every fifteen minutes. This is possible due to new Smart Parks technology providing Rangers with GPS-locations four times an hour. With this information, elephants can be better protected.

At IRNAS we have set up a unique combination of development and manufacturing capabilities, empowering our rapid innovation and taking customers from ideas to batches of devices in minimal time. Having our own infrastructure gives us a high level of control over our in-house manufacturing and assembly processes. With that, it comes also making sure that design decisions are based upon manufacturing and assembly capabilities.

Image for post
Image for post

In this blog post, we will talk about using adhesives and glues, which we’ve been avoiding to some extent in the past for two main reasons.

First, glueing is a messy, time-consuming process. Of course one would happily apply glue on one piece of hardware as a one-off solution, however we persistently tried to steer away from using glue when it comes to producing 20, 100 or 1000 pieces, which is a production scale will typically do in-house. …


Author: Vid Rajtmajer, student intern at IRNAS

Image for post
Image for post

Introduction

Fault-tolerant solutions result in happy users, which remain oblivious to certain system failures if they are recovered in time. At IRNAS we rapidly develop advanced applied solutions and often design complex systems with rapid innovation. In experimental phases of the projects, these get deployed to various locations where they are stress tested and also user stories are yet to be found and defined. IRNAS team is usually the first observer of a user actually using the device for the first time and this is where we often learn the most. With innovation, it is impossible to predict all things that can go wrong, and that’s something we have to be content with. …


Author: Marko Sagadin, student intern at IRNAS

Image for post
Image for post

Welcome to the second article about running machine learning algorithms on microcontrollers.
In the previous article, we have created and trained a simple Keras model that was able to classify 3 different classes from CIFAR-10 dataset.
We looked at how to prepare that model for running on a microcontroller, quantized it and saved it to disk as a C file.

It is now time to look at how to setup the development environment and how to run inference on an actual microcontroller.

For this we will need two things to follow along. First, we will need to clone MicroMl project, which will provide us with the example code. …


Author: Marko Sagadin, student intern at IRNAS

Image for post
Image for post

In the past few years, there has been slow but constant push to run neural networks on small embedded devices. The most obvious benefit this feature can bring is not needing to send data over the radio to a server for computation. We can do everything we need to do on the device itself, thus saving the energy that would be otherwise used for wireless transmission.

We have already seen examples of products such as Amazon’s Echo or Google’s Google Home that listen all the time for a specific keyword and only send your data to the server after hearing that word. A specific chip was required for such operation in the past, but now, thanks to TensorFlow Lite for Microcontrollers, we can flash a pre-trained neural network model to a common microcontroller and run it on the device. …


Author: Eva Černčič

The innovation company IRNAS, providing the smart IoT solutions for industrial applications and the cutting-edge technology manufacturer of overvoltage protective and EMI suppression components, Bourns, did join forces on the NexGenHVEC project to accelerate development and advance the production line by implementing data collection and machine learning technologies.

Image for post
Image for post

The main aim of the NexGenHVEC project is to advance the industry sector by utilizing cutting edge technology and modern research capabilities. The research goal is to improve the thermal stability of hybrid electronic systems and components (varistors and capacitors) through the development of advanced materials and improved manufacturing practices. …

About

Institute IRNAS

We are applying today’s knowledge to create systems for an open future.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store