Real-time IoT data visualisation is difficult when you’ve got complex mechanical setups. But what if you could see that system in the cloud as if you were in the room, and be able to predict when it’s going to go wrong?
In ipXchange’s second interview on the Supermicro booth at Embedded World 2024 – brought to you in partnership with Memorysolution – we chat with Ben from Kaa IoT about a very interesting demonstration of digital twinning.
As Ben explains, Kaa IoT integrated its enterprise IoT platform with Supermicro’s processing hardware so that the robotic arms you see, and the coloured blocks that they are interacting with, are digitally represented within the cloud. This is all done using live sensor data from the arms, with AI image inferencing – A.K.A. identifying the blocks – done at the edge. Surprisingly for some, this is done without the use of a GPU.
In this demo, a gold block indicates a defective item in a production line, and the system has been programmed to deal with this by taking it out of the loop. An additional camera running an AI model on the same system is used to identify faces, simulating a security system that can inform the setup when an unauthorised user is by the machine. Alternatively, this could be trained to identify whether people around the machine are wearing the correct safety equipment.
A key benefit of such a well-characterised system – enough to create a digital twin – is for predictive and prescriptive maintenance, since the system can examine everything that happened to it leading up to a break.
It can also be used to improve the performance of the system by examining different ways of operating and figuring out what elements of each worked most efficiently.
While this system is obviously mirroring a factory automation application, Ben highlights that any low-latency monitoring or security application would benefit from this type of setup. This is especially true if the raw data needs to be kept off the cloud for security or privacy reasons. Examples include warehousing automation, security access and tracking, or patient monitoring.
Kaa IoT streamlines development of industrial-grade cloud platforms that can be tailored exactly to your requirements, with the ability to connect a broad range of communication protocols to your service.
In terms of the hardware, Supermicro provides various compute solutions geared towards running AI workloads at the edge. ipXchange’s favourite is the SuperServer E100-12T-H, which features the Hailo-8 AI Accelerator with 26 TOPS of AI performance and a power consumption of just 3 W. Head to the board page via the link below to apply to evaluate this technology today.
Keep designing!
Enjoy powerful AI solutions at the edge? Check out these other interviews from Embedded World 2024, including the others brought to you by Memorysolution:
Fanless industrial computers for 24/7 AI tasks
Hailo’s M.2 card does gen-AI at the edge at 3.5 W
Machine vision with the Hailo-8 AI accelerator
Launching Exascend’s PR4 radiation-hardened SSDs