IoT news of the week for Feb. 12, 2021

COVID-19 is a great IoT sales pitch: COVID-19 has helped create the ultimate use case for the internet of things, and companies are not shy about bringing these use cases to the forefront in an effort to sell their services. Whether it’s an organization that makes software to track machine health and pitches that as a way to share factory data with remote workers; a telco trying to sell communications networks in a factory, on the road, or in the middle of nowhere; an org that has sensors and a cold-chain monitoring system; or one that is simply focused on telemedicine, COVID has made IoT necessary and compelling. The latest example comes from AT&T which is pitching its network and software as necessary tools for vaccine tracking. (AT&T)

IBM signs deal with Foghorn Systems: Foghorn, which provides edge computing software and has a deal with Google to connect manufacturing clients’ data back to Google’s cloud, has signed a similar deal with IBM to bring edge analytics and data from client sites back to IBM’s cloud. The idea is that Foghorn will provide the analytics and device data integration coming off machines and the plant’s OT network, while IBM will provide the infrastructure needed to parse data in the cloud. (Foghorn)

It’s time to digitize the railroad industry: Since I’m in the U.S., I don’t spend a lot of time thinking about rail, even though it’s clearly an integral part of global travel and an even larger component of freight movement around the globe. That is why I was so stoked to find this deep dive into what matters for companies trying to connect different parts of the rail industry. I learned that railways need explosion-rated devices, was introduced to the components rail operators are interested in monitoring, and got an overview of several organizations that are trying to set standards and protocols across the industry. If you want to sell a sensor or software to rail operators, read this first. Yes, it is written by the founder of a company trying to build hardware and software to digitize rail operations, but it offers a ton of knowledge to those of us who aren’t familiar with the opportunity. (Railway Age)

Google is filing a bunch of patents for stuff we’ve talked about forever: I know that with patents, the devil is in the details, but the patent filings in this story are for things that don’t seem novel enough to merit one. Among them is a system of sensors from Google that shares information regarding how a senior moves about their home, issuing alerts if the person in question doesn’t move around in the morning, etc. However, I’m excited to see Google thinking about how to create retrofit sensors that could stick to clothing, toys, or whatever. That said, I have been seeing variations on these sorts of connected sensor tags since as far back as 2008, when Alcatel-Lucent (whoa, blast from the past) shared a similar vision. Pretty sure I also saw something similar in the early aughts, at IBM’s smart house lab. (Protocol)

Are you thinking about securing your ML models? This article which is based on a Microsoft presentation is all about the need for companies to start thinking about how to secure their machine learning models. The idea is not so much that a hacker could insert inaccurate training data into a model, and thus render it confused or bad, but more that companies should understand all of the potential ways outsiders could tamper with their operations. (The Register)

Uptake acquires ShookIOT: I haven’t heard much from Uptake in a few years, but it’s still trucking along, and has just purchased a Canadian startup called ShookIOT, which specializes in machine data ingestion. Uptake was one of the early companies involved in industrial IoT, building software that helped pull data from machines and building cloud connectors. (Uptake)

Amazon Alexa’s new talent is sharing songs between friends: This is a neat little trick from Alexa. You can now send a song to someone’s Alexa if they are in your Alexa contact list. This feels cutesy and somewhat random, although music is a nice way to connect with a far-away friend or show support to a family member. But Amazon says this feature was “just the beginning” for the feature, which has me thinking about how Alexa could become the gatekeeper to my closest friends and family members in a voice-centric social network. (Amazon)

Kevin is switching to HomeKit to protect his privacy: My colleague Kevin has been on a tear in the last few months, dumping services that suck up his data but fail to provide enough consent or value. As such, he’s now trying on the HomeKit ecosystem by replacing some of his Google gear and downloading Homebridge to make the connection between his Apple stuff and his non-HomeKit products. (StaceyonIoT)

This is how the robots get dogs on their side: This research is kind of fun, but it also shows the great lengths companies and researchers have to go to in order to train artificial intelligence systems to recognize basic movements. In the experiment, researchers taught a computer how to understand if a dog is sitting, lying down, or standing. When the computer identified that the dog had performed the requested task correctly, it would deliver a treat. The researchers also said they might use the algorithm for a future dog training product. I don’t know about you, but I don’t want my dog colluding with Spot when the robots decide to take over. (VentureBeat)