IoT news of the week for Oct. 25, 2019

OCF just subsumed another standards organization to bring about interoperable smart buildings: The Open Connectivity Foundation, a group trying to promote an interoperable standard that would allow connected devices to share identifying and functional information amongst themselves, has integrated with the Fairhair Alliance. The Fairhair Alliance works with lighting, building automation, and IT companies to develop a secure onboarding framework for lighting and building control in commercial buildings. (OCF)

Foundries.io raises $3.5M: Foundries.io, a company that is building an OS as a service for the internet of things, has raised a first round of funding from Crane Venture Partners and BackedVC. I profiled the startup back in July. Since then it has launched Foundries Factory, which is a service designed to provide an over-the-air, updatable operating system for IoT devices that loads from the secure enclave on a chip. This helps keeps edge devices secure and up to date. (EENews Europe)

Computer vision knows if you’re wearing your hard hat: Flutura, a company that provides analytics for industrial IoT environments, has launched a use case-specific video analytics product. The product runs on top of Flutura’s existing software and is pre-trained to detect hard hat violations, objects, people or objects entering restricted zones, and welding defects in X-ray images. The software uses existing CCTV feeds for detection and links findings back to diagnostic and other plant data to help show how safety violations can relate to downtime. (Flutura)

That’s a lot of smart fridges: The connected fridge used to be a staple of joking coverage by the IoT world. But now its day has come. According to data from Futuresource Consulting (and the conversations I have with makers of home appliances), the share of connected fridges, washing machines, and dryers is growing. Futuresource estimates that annual shipments of connected fridges and laundry machines will double between 2018 and 2020, to reach 33 million units worldwide. That’s 10% of the forecasted 326 million shipped fridges and laundry machines. (Futuresource Consulting)

A good argument for edge computing: This article opens with the suggestion that billions of sensors and the computers to process their data will require huge amounts of energy and drive climate change, but quickly delves into the impossible amounts of video data generated when we add computer vision to homes, stores, and factories. ARM SVP Drew Henry then talks about energy-sipping chips and energy-harvesting chips in a vague way in response to a question he’s asked about energy use. (While I believe energy-harvesting tech is essential to IoT, I don’t see ARM making any huge bets on it, nor will it address the enormous processing requirements needed to train machine learning models.) In short, the IoT won’t work without energy-harvesting sensors. It’s a good interview, even if Henry punts on some of the hard questions. (VentureBeat)

IoT is set to change the construction industry: Disperse, a company using video sensing to improve efficiencies at construction sites, has raised $15 million in venture funding. The Series A round led by Northzone will help the UK-based startup expand in North America. Disperse uses sensors and cameras to provide insights associated with construction delays, materials waste, and more. It’s part of a wave of startups trying to bring tech into construction. (Disperse)

An update on Bose’s health ambitions: Bose, long associated with high-quality sound in consumer audio equipment, said it would move into health wearables about 18 months ago. This story provides a lot of context on its effort to build hearing aids and sleep aids, and offers perspective on the challenges of moving from a consumer-oriented business to a medical-device business. Since I’m obsessed with how connected hearing aids and a change in legislation will allow hearing aids to be sold over the counter, I am keen to see what Bose comes up with here. (Fast Company)

Understanding what it takes to build ML models for Google’s Project Soli: This is deeply nerdy, but a fascinating (and old) paper that I’m highlighting because Soli is in the news. It discusses the challenges associated with turning disruptions in RF signals into meaningful gestures using a variety of neural networks. The challenges range from figuring out motion as opposed to a static image, to the levels of uncertainty that a model should allow in taking an action based on the gesture. It also illustrates how computationally significant such models are. There’s a reason Soli is so limited today. (SIGCHI)

More privacy implications of facial recognition software: Physicians were recently surprised to learn that the medical records of patients that include MRI scans can violate the rules set around patient privacy. A researcher discovered that the faces scanned as part of an MRI scan can be accurately identified in 70 out of 84 faces studied. Which means that despite stripping a medical record of personally identifiable information, a patient’s identity could still be discovered. Which means that perhaps MRI scans shouldn’t be shared. (Medgadget)

Sierra Wireless has created a new industrial IoT platform: Sure, why not add to the mix of IIoT platforms? Sierra Wireless has launched a new IIoT platform known as Octave, and its aim is actually somewhat noteworthy: to help customers figure out the infrastructure needed to solve their problems, and link easily to a variety of platforms that will comprise the overarching “solution.” Think of it like an IIoT wizard, helping lead customers through a complex and technical procurement process. (IoT World Today)