TDK's new low power smart sensors. Mentions their engineers "built an alternative in which a machine learning algorithm could recognize motion patterns at the sensor level, to determine whether further data processing is required. In that way a device could be ultra-low power."
Seems a bit fiddly and not very versatile IMO.
New Technology Aims at Easy, Low Power IoT for Manufacturing
BY CLAIRE SWEDBERG
The new BLE-enabled module and software platform from TDK are part of the company’s drive to make sensor-based equipment monitoring with AI
Jan 29, 2024For those in industrial settings, digital management of machine operation may be easier, with a new solution and sensor from electronic components company
TDK corporation.
The company is offering its Smart Sensing Platform to enable faster deployment for Internet of Things (IoT) devices or other wireless sensing technology.
The company has released its Bluetooth mesh connected I3 device—with AI on-the-edge capability for the industrial IoT. It captures sensor data and shares it, as well as inference data, wirelessly, but requires low power based on its ability to send only relevant data, when it’s needed. The company demonstrated the new products at CES.
Making Sensors Smarter
TDK’s primary businesses are divided into three main groups: ICT (information and communication technology), automotive, and industrial and energy. Within these groups, one of the technology solution consists of IoT-based sensors that measure conditions and share that data via wired or wireless connections.
“We've been [primarily] focused on making our sensors smarter,” says Jim Tran, TDK USA’s general manager, and part of that effort is edge computing. “When we say edge, [in this case] we really mean at the very edge —on the sensor itself.”
As a result, the company has built its SmartEdge machine learning algorithm that links to a motion sensor within a device so that it can detect a motion and transmit accordingly. When the device is stationary, it can remain dormant.
Low Power I3 Sensor
Tran notes most wearable technology comes with a form of motion sensing. Traditionally, devices process the motion data on a CPU or other dedicated hardware to identify what that motion means.
TDK engineers built an alternative in which a machine learning algorithm could recognize motion patterns at the sensor level, to determine whether further data processing is required. In that way a device could be ultra-low power.
The company refers to the resulting technology as “sipping currents” or micro droplets of energy that are required to track conditions with the new devices.
The I3 module, measuring about the size of a quarter, is a resulting product for electronic device developers focused on measuring machine health. It comes with built-in BLE beacon for industrial mesh networks.
SmartEdge Algorithm
The overall solution that enables the latest, low energy, IoT deployments is TDK’s Smart Sensing Platform, featuring sensors and software with edge AI, connectivity and cloud computing. The goal is to make deployments easier and more seamless, with always-on interactive apps and services. The solution leverages the company’s SmartEdge AI algorithms.
The algorithms enable users to run machine learning at the edge, using select sensor features such as vibration profile or temperature requirements to identify what is taking place and when data needs to be forwarded to the server.
Users could apply the TDK I3 or other IoT sensor devices on machinery in a factory or industrial site, which then begin tracking data about the sound, vibration or temperature being emitted by each machine. The devices could then use a Bluetooth mesh network to forward the data to a Wi-Fi access point when necessary.
However, the system is intended to send only relevant data to the cloud. TDK’s smart edge platform infers specific conditions before transmitting that data.
Ease of Integration With Less Engineering
For developers building AI intelligence into a sensor, the process requires several steps, says Tran.
“You need to be able to make that algorithm super small for the tiniest memory footprint, for cost and latency,” he said.
He adds that the next step is having AI engineers available to write an algorithm for each deployment, and in some cases, for each kind of sensor device, or equipment that the device is monitoring.
To that end, TDK recently acquired Carnegie Mellon spinoff company Qeexo which created the developer tools that simplifies the process.
A Set of Machine Learning Algorithms
In most cases engineers or developers would need to conduct modeling and coding using C or C++ code in
Python—leveraging trained domain experts that understand that data so they can label it. AI engineers can bypass several processes with this technology, however.
Using the TDK solution, a sensor can employ any of 18 machine-learning algorithms designed for edge sensing.
Developers select the algorithm, convert it to machine code and then download it to a sensor. They then apply the sensor to the machine to begin monitoring the data.
Transforming Manufacturing to Industrial 4.0
“We believe that this type of solution is really needed in order to scale for industrial 4.0,” says Tran.
The technology is being adopted by companies such as factories that use smart sensors, as well as by solution providers that license TDK’s tool and create their own edge AI products.
“We're focused on using this tool combined with our devices like the I3 and to really help factories transform themselves into the digital world,” Tran says, adding for him, it’s a way to democratize IoT and AI solutions.
The goal is to make it possible for companies to deploy a solution without hiring outside engineers.
Do-it-Yourself Development
In that way, some companies could develop their IoT solution without needing to hire outside expertise. “They can do it themselves which simplifies everything dramatically,” said David Almoslino, corporate marketing SVP.
Last year,
Procter and Gamble announced the use of this tool for their product development which the company says reduces the development time of their AI algorithms. They have not shared how they use the technology specifically.
The technology helps companies better manage conditions even at customer sites. For instance, firms that sell or lease equipment used at manufacturing sites can identify problems that could arise. In the case of a break-down, they would have access to troubleshooting data even before service personnel are onsite.
Key Takeaways:
- TDK releases new sensor and software to enable easier, low energy IoT for equipment monitoring.
- Companies such as Procter and Gamble are using TDK’s solution for AI on the edge functionality.
The new BLE-enabled module and software platform from TDK Corporation are part of the company’s drive to make sensor-based equipment monitoring with AI, available to more users.
www.rfidjournal.com
for Gregor Lenz’s start-up Neurobus… And one more from another Brainchip employee - Gilles Bézard (remember my post yesterday? Gilles and Gregor are both on the scientific committee for the inaugural SPAICE 2024 conference). Oh, hang on, and Wouter Benoot, CTO of the Belgian SpaceTech company EDGX likes this post as well! (
