New technology aims for simple, low-power IoT manufacturing – (WIW)
New technology aims for simple, low-power IoT manufacturing – (WIW)
TDK’s new BLE-enabled module and software platform is part of the company’s efforts to revolutionize sensor-based devices with AI.
The 29- January 2024 For industrial users, settings, digital control of machine operation can be made easier with a new solution and sensor from electronic component company TDK Oyj.
The company offers a Smart Sensing Platform that enables faster deployment of Internet of Things (IoT) devices or other wireless sensing technology.
The company launched its Bluetooth-connected I3 device with cutting-edge AI capability for Industrial Internet of Things. It captures sensor data and shares and infers data wirelessly, but requires little power because it can send important data only when it is needed. The company presented new products at CES.
Smarter than sensors
TDK’s main activities are divided into three main groups: ICT (information and communication technology), automotive industry and industry and energy. In these groups, one of the technological solutions consists of IoT-based sensors that measure conditions and share data through wired or wireless connections.
“We’re focused [primarily] on making our sensors smarter,” says Jim Tran, CEO of TDK USA, and part of that effort is edge computing. “When we say edge, [in this case] we really mean the edge – the sensor itself.”
As a result, the company created its own SmartEdge machine learning algorithm that connects to the motion sensor inside the device to detect movement and send it accordingly. When the device is stationary, it can remain in sleep mode.
Low Power I3 Sensor
Tran notes that most wearable technology includes a motion sensor. Traditionally, devices process motion data with a processor or other separate hardware to determine what the motion means.
TDK engineers have created an alternative where a machine learning algorithm can detect motion patterns at the sensor level to determine if additional data processing is required. In this way, the device can be very low power.
The company calls the resulting technology “absorption currents”, or micro-droplets of energy, needed to monitor conditions in new devices.
The I3 module, approximately the size of a quarter, is a performance product for electronic device developers focused on machine condition measurement. It has a built-in BLE beacon for industrial networks.
SmartEdge Algorithm
A complete solution enabling the latest low-power IoT deployments is the TDK Smart Sensing Platform, which includes sensors and software with AI, connectivity and cloud computing. The goal is to make deployment easier and smoother with always-on interactive apps and services. The solution uses the company’s SmartEdge AI algorithms.
Algorithms allow users to perform machine learning at the edge using specific sensor characteristics, such as vibration profile or temperature requirements, to determine what is happening and when data should be sent to the server.
Users can connect TDK I3 or other IoT sensors to machines in a factory or industrial area, which will then start monitoring the sound, vibration or temperature data emitted by each machine. Devices can then use the Bluetooth network to transfer data to a Wi-Fi hotspot when needed.
However, the system should only send relevant data to the cloud. TDK’s smart edge platform infers certain conditions before sending data.
Easy integration with a smaller design
For developers building AI into a sensor, the process requires several steps, says Tran.
“You have to be able to make this algorithm very small for the least amount of memory, cost and latency,” he said.
He adds that the next step is for AI engineers to write an algorithm for each deployment and, in some cases, each sensor device or device monitored by the device.
To that end, TDK recently acquired Qexo, a Carnegie Mellon spin-off company that created software tools to simplify the process.
A set of machine learning algorithms
In most cases, engineers or programmers must model and code in C or C++ code in Python, using trained domain experts who understand the data to make these annotations. However, AI engineers can bypass several processes with this technique. When using the TDK solution, the sensor can use any of 18 machine learning algorithms designed for edge detection.
Programmers choose an algorithm, convert it to machine code, and then upload it to the sensor. They then connect the sensor to the machine and start tracking the data.
Transforming Manufacturing for Industry 4.0
“We believe this type of solution is really necessary to scale Industry 4.0,” says Tran. Companies that have adopted the technology, such as factories using smart sensors, as well as solution providers that license the TDK tool and create their own extreme AI products.
“We’re focused on using this tool with our devices like the I3 and making factories truly digital,” says Tran, adding that this is a way to democratize IoT and AI solutions.
The goal is for companies to implement the solution without hiring external engineers.
Self-development
This is how some companies can develop their Internet of Things solution without external experts. “They can do it themselves, which simplifies everything dramatically,” said David Almoslino, the company’s director of marketing.
Last year, Procter and Gamble announced that they are using the tool for their product development, which the company says will reduce the development time of AI algorithms. They did not say how they will use the technology specifically.
Technology helps companies better manage conditions at customers’ sites as well. For example, companies that sell or lease equipment used in production facilities can identify potential problems. In the event of a breakdown, they have access to troubleshooting before service personnel arrive.