TinyML, essentially a way of optimizing and reducing machine learning to work on small form factor, low power, and memory edge devices, will be key as more organizations seek to develop critical data analytics and deep learning. Shifting functions from the cloud to the cloud takes the edge.
This week, Imagimob, a Sweden-based software-as-a-service company pushing edge AI applications, said it had successfully demonstrated how TinyML could be used in agricultural applications what farms and “agritech” companies do otherwise save high investments in the cloud. based machine learning approaches.
“If you can move an application from the cloud to the edge, you can make significant savings through lower cloud operations costs,” said Anders Hardebring, CEO and co-founder of Imagimob, who co-founded. spoke Violent electronics by email. âAnd also reduced communication costs. You can save a lot of money with predictive maintenance. “
The most recent demonstration, which was part of a three-year project that ended this month, took place in a national park near Pisa, Italy, where Imagimob worked with partner STMicroelectronics, as well as sensor device maker Bosch and semiconductor company Expressif.
A blog post by Hardebring explains: âWe used two different tractors for the demonstrator, a modern tractor from the German SDF group and an older, less ‘networked’ tractor from the American company Deere & Co classify so that the farmer’s end-user can manage the tractor fleet through a Grafana dashboard; where they are, their operating mode in near real time and statistics for use for planning and maintenance. “
He added: âIn effect, we installed a Dialog IoT Kit (Bosch Sensor) device, an Android phone in a dashboard mount. The data can either be flagged in real time by the driver / operator or in post-production by displaying the recorded smartphone video stream. “
Hardebring continued: âWith our Imagimob AI software, a number of neural networks were trained and provided on the device along with sensors, batteries and a LoRa radio. The end result enables the farmer to monitor mobile assets using an extensive data set from the accelerometer and gyroscope and regularly send the end result over the LoRa network for tracking in near real time. “
He said Violent electronics that the combination of running TinyML applications on edge devices using LoRa as a wireless network is a “powerful combination” of tools with common characteristics in that they are inexpensive and use little power, but still perform well.
Hardebring said the company used the same technology and tools to set up a farm animal tracking application that uses smart collars on cattle to monitor their health.
Upon completion of the project earlier this month, Imagimob will speak to “a number of parties who want to take the concept to the next level,” he said, adding that Imagimob is also working in other areas of Industry 4.0 in addition to Agritech in the automotive, consumer electronics and other fields .
As TinyML continued to evolve, Hardebring said that because of its low cost and low power consumption, it would be the most common candidate for greenfield deployment options where those requirements and low storage footprint are high on the list. This could, for example, be carried over to use cases in remote or rural areas in developing countries, where other, more expensive IoT options are simply not considered.
RELATED: Qeexo Adds AutoML to STMicro MLC Sensors to Accelerate tinyML and IIoT development