I was at the Parks Connections event that covers the smart home this week, so I share a few thoughts on what’s holding back adoption and how to think about using AI to create a smart home. From there, Kevin talks about the new meeting function offered by Alexa and we add nuance to the debate over Amazon selling facial recognition software to police. We then dig into some additional doubts about the new Wi-Fi EasyMesh standard, cover Comcast expanding the places it offers new Wi-Fi pods, discuss funding for a smart light switch company and new Arduino boards. For the more industrial and maker minded, we talk about Ayla adding Google Cloud as a hosting option and Kevin shares how we put our IoT hotline into the cloud. Finally, we answer a question about getting different bulbs to work together before switching to our guest.
This week’s guest is Mitch Bowling, the CEO of Sears Home Services, who gives me the answer to what Sears plans to do with its acquisition of Wally sensor business back in 2015. I have been wondering what happened to Wally inside Sears for years. He also discusses how Sears can use IoT to make appliance repair better and the plans to add smart home installation services. Enjoy the show.
Hosts: Stacey Higginbotham and Kevin Tofel
Guests: Mitch Bowling, CEO of Sears Home Services
Sponsors: MachineQ and Bosch
- Device interoperability is a huge challenge for the smart home
- The fuss over computer vision is just beginning
- What can Sears Home Services do with IoT?
- The smart appliances are coming!
- The installer will see you now
Podcast: Play in new window | Download | Embed
Subscribe: RSS
Ep: 165. Great show as always.
First time, long time.
Your discussion of Amazon’s facial recognition software for the police made me think about the rollout of fingerprinting techniques for police work back in the 20th century. There must have been a specialty of people who could “analyse” fingerprint images and then “match” them to the suspect. Surely at some point a computer scanning system must have been developed so the fingerprints can be scanned automatically.
Some years ago I worked with some primitive fingerprint systems for door access. They used a stored image of the fingerprint and the device would analyse the image for unique characteristics, usually 8 points, which gets hashed into a 256 digit number, the key. The concern quickly moved away from counterfeit using fingerprint molds to just entering the number in directly, either hacking the ID key db or brute force.
Anyway I am pretty sure the fingerprint sensors on mobile devices use a similar though much more advanced system.
I suspect that the facial recognition software works along this same line of unique features that generate a calculated number so that the main algorithm can sort and search the db quickly.
This makes the facial recognition system much more useful for nationwide searches of mugshots for people with arrest warrants.
For example if I am wanted in New Jersey for holding up a casino with my 7 pals and I get arrested for “making it rain in the club” in North Carolina my current mugshot should generate a fast and accurate ID that could be checked against the warrant in NJ. In my mind this is really just an extension of automatic fingerprint reading.
I don’t know whether the fingerprint scanner in my phone sends that image & key back to somebody else. I wonder about the facial unlock tool as well. Are the image and the generated ID key stored and accessible on the phone? Surely they must be. Is that being sent to somebody too?
It would be possible, though improbable, to capture and match fingerprint id, facial id, IP, mobile MAC address along with SSN, drivers lic # and the usual ID items. By linking the digital identity to the physical ID, lets not forget location data, the surveillance apparatus is nearly complete.
(Sorry for the long ramble. Love the show.) /Rick in Raleigh