Imagimob agritech demo shows off TinyML's promise for edge IoT applications

TinyML, essentially a way of optimizing and shrinking machine learning to work on small form factor, low-power, low-memory edge devices, is going to be key as more companies look to move important data analysis and deep learning functions from the cloud to the edge.

This week, Imagimob, a Sweden-based software-as-a-service company forced on edge AI applications, said it successfully demonstrated how TinyML could be used in agricultural applications which could save farms and “agritech” firms otherwise hefty investments in cloud-based machine learning approaches.

“If you can move an application from the cloud to the edge, you can make significant savings in reduced costs for cloud operations,” said Anders Hardebring, CEO and co-founder of Imagimob, who spoke with Fierce Electronics via email. “And also reduced communication costs. In predictive maintenance you can save a lot of money.”

The recent demonstration, which was part of a three-year project ending this month, took place in a national park area near Pisa, Italy, where Imagimob worked with partner STMicroelectronics, as well as sensor device supplier Bosch and semiconductor firm Expressif.

A blog post from Hardebring explained, “We used two different tractors for the demonstrator, one modern tractor from German SDF-Group and an older, less ‘connected’ tractor from the American company Deere & Co. By collecting sensor data on movement, we trained a neural network to classify the operational conditions, allowing the farmer end-user to manage the fleet of tractors via a Grafana dashboard; where they are located, their operational mode in near-real time and statistics on usage for planning and maintenance.”

He added, “Practically, we installed a Dialog IoT Kit (Bosch sensor) device, an Android phone in a dashboard holder. Data can be labeled either in real-time by the driver/operator or in post production by viewing the captured smartphone video stream.”

Hardebring further stated, “Using our Imagimob AI software, a number of neural networks were trained, and deployed on-device together with sensors, batteries and a LoRa radio. The end-result allows the farmer to monitor mobile assets using a rich data set from accelerometer and gyroscope, sending the end-result over LoRa networking periodically for tracking in near-real-time.”

He told Fierce Electronics that the combination of running TinyML applications on edge devices that use LoRa as the wireless network is a “powerful combination” of tools that share common attributes, in that they are low-cost and consume little power, but still perform at a high level.

Hardebring said the company also used the same technology and tools to set up a livestock tracking application in which smart collars on cattle to monitor their health.

With the project wrapping up this month, Imagimob is talking to “a number of parties that want to take the concept to the next level,” he said, adding that in addition to agritech, Imagimob also is working in automotive, consumer electronics and on other Industry 4.0 opportunities.

As TinyML continues to develop, Hardebring said its low-cost and low-power properties will find it most often as a candidate for greenfield deployment opportunities where these requirements, as well as low memory draw, are at the top of the list. For example, that could translate to use cases in remote or rural areas in developing countries where other, more expensive IoT options just aren't an option to be considered.

RELATED: Qeexo adds AutoML to STMicro MLC sensors to speed tinyML, IIoT development