All News | Boards | Chips | Devices | Software | LinuxDevices.com Archive | About | Contact | Subscribe
Follow HackerBoards:
Twitter Google+ Facebook RSS feed
(get email updates)
Please whitelist HackerBoards.com in your ad blocker. Without ads from our sponsors, we cannot continue publishing this site. Thanks :-)

Linux-based smart home hubs advance into AI

Oct 13, 2016 — by Eric Brown — 3,390 views

Linux-based smart home hubs continue to rule the home automation world, with the latest products integrating voice assistants and AI analytics.

 
 

An Open Source Perspective on the Internet of Things
Part 4: Linux-based Smart Home Hubs Advance into AI

 

Industrial, rather than home, applications will likely dominate the Internet of Things (IoT) market in the years to come. Yet, in the early going, the home automation market has had the greatest visibility. And it hasn’t always been pretty.

Despite steady growth, retail sales have yet to achieve inflated expectations. Too many companies promised and failed to deliver interoperability with a growing catalog of often buggy smart home products. The lack of essential applications, complex installation, and in many cases, high prices, have also conspired against the segment.



Google Home

Yet the smart home segment appears to be rebounding with the help of maturing technology and IoT interoperability standards. There is particular interest in connecting voice-enabled AI assistants with the smart home in products such as Amazon’s Echo. Google recently announced Google Home, which aims to be a major competitor to Alexa by combining its Google Assistant voice agent with a version of its Linux-based Chromecast stack. These highly publicized devices are being joined by open source Linux smart home voice agents like Mycroft, Silk, and ZOE (see farther below).


Mycroft

Most home automation systems run on Linux hubs or revolve around higher-end Linux-based smart devices like the Nest Thermostat. Many of these have fairly open APIs for app development, although the hardware is usually locked down.

As we explored over the last two weeks, open source hackers can also build fully open smart home systems using personal computers or Linux hacker boards running open source IoT stacks. Developers are also hacking commercial hubs to load open software.

A few of the products in Linux.com’s August 2014 roundup of Linux-based IoT smart home hubs have been discontinued. These include the innovative, open source Ninja Sphere. Fortunately, Ninja Blocks was able to send units or refunds to backers before closing shop, and the software is free for continued experimentation.

This summer, Staples scrapped its Staples Connect system. The Connect devices will continue to be distributed by Z-Wave Products, including the Linux-based, Alljoyn-compliant D-Link Staples Connected Home Hub, even if D-Link is no longer promoting the hub with its other smart home cameras and devices.

One of the big early players — Revolv — was acquired by Alphabet-owned Nest. In April Revolv was discontinued, demonstrating the risks involved with buying into a home automation system.



SmartThings Gen 2 hub with peripherals

Samsung’s acquisition of SmartThings has been more profitable. Not only has Samsung moved SmartThings to a more open, Linux-based platform, but it’s expanding with new devices and third-party device support. SmartThings remains one of the market leaders.


Wink Hub 2.0

The heavily promoted Wink system, which was backed by GE and sold at Home Depot, suffered from negative reviews, and its owner, Quirky, entered bankruptcy. Wink was acquired by Flex, however, and last week released a much improved, $99 Wink Hub 2.0. Wink also announced a major new distributor with Walmart.

Wink confirmed that the new hub continues to run “a custom-built Linux OS,” supported with an updated, open Wink API based on RESTful. There’s also wide-ranging, third-party device support. The Wink hub is proprietary, but has been rooted and loaded with OpenHAB, as well as the similarly open source Home Assistant.

Like Staples Connect and Wink, Lowe’s Iris has survived thanks to a strong retail presence. This year, Iris 2.0 made substantial improvements to its lagging technology, including Microsoft Azure integration and AllJoyn and Amazon Echo compatibility.

Some home automation players, such as Insteon, continue to compete with a closed system. Similarly, the Lutron Smart Bridge is devoted to Lutron products using its proprietary ClearConnect wireless protocol, such as its Caseta dimmers. There’s also LG, with its proprietary SmarThinq, among others. The most popular proprietary system is Apple’s HomeKit, which draws upon a devoted customer base.

Some smart home systems, such as Vera, support just about any product with Z-Wave — a more proprietary technology than the similarly short-range ZigBee. The Google-backed Thread Group is aiming to replace both standards with its IPv6-infused, mesh-networking Thread protocol based on the similarly 802.15.4-based 6LoWPAN. For an open source software spin on Z-Wave, IoT hackers can turn to Open-ZWave, which is building an open source software library for Z-Wave PC controllers.

Newer IoT systems tend to support multiple wireless protocols, and have more open APIs, even if they are rarely fully open source. With the growing availability and sophistication of smart lighting products such as Philips Hue and Lifx, or smart locks from Kwikset, August, Schlage, Yale, hub/stack vendors do not need to field their own device competitors. Instead, they can focus on advanced software and hubs that are compliant with major smart devices and open source frameworks like The Linux Foundation hosted projects AllJoyn and IoTivity.



WigWag Relay

One of the bigger open source success stories in the smart home business is WigWag, which offers a hackable, Linux-based Relay hub along with a sampling of homegrown sensors, presence tags, and lighting devices. WigWag depends primarily, however, on third-party smart devices compatible with AllJoyn, IoTivity, and Thread. Another open source contender is Webee, which offers a self-learning hub that lets you use your TV as an interface.


Nest Thermostat

Consumers tend to start off in the home IoT market by purchasing a single-purpose smart device such as a thermostat or smart lock, controlled via a smartphone app. The most successful product line without a central multipurpose hub is Nest Labs’ Nest Thermostat. Despite a leadership shakeup this year, the Alphabet-owned Nest continues as a market leader. Other popular hub/endpoint hybrids remaining from our 2014 list include Belkin’s WeMo and iControl’s Piper home security camera.

 
Voice assistants and analytics

Although some home automation systems feature wall-mounted touchscreens, the typical interface has been the smartphone app. Unfortunately, smartphones are overloaded with other duties, and have a tendency to get left behind in the other room. Even if you have a unified app for all your smart devices, digging into the menus can take time.

Voice response systems are increasingly seen as a more user-friendly smart home interface that can also extend to multimedia access and general web queries. A growing number of vendors including SmartThings and WeMo, as well as old-time home automation firms like Creston and Control4, are adding support for the popular Amazon Echo speaker/IoT hub. The main draw of the Echo is the Alexa voice agent, which enables the Echo to act as a voice-activated smart home hub.

The Echo is not open source, but last year Amazon opened up the Alexa SDK, thereby greatly extending its reach. Integrating with the Echo is certainly easier than hooking up with Apple’s proprietary, HomeKit, which remain popular in part due to its Siri voice integration.

Google and Microsoft are readying their own voice agents for smart home applications to compete with Alexa and Siri. Google Home offers a voice-enabled smart speaker, and Google is said to be encouraging other tech firms to build their own Google Home compatible devices built around its Google Cast and Google Assistant technologies.

Earlier this year, Microsoft announced that its Cortana voice assistant for Windows 10 would expand beyond its AllJoyn compatibility to support IoTivity. A Cortana-ready smart home product should follow in 2017.


ZOE

Newcomers are challenging these tech giants with their own voice-enabled products. While the major voice agents are all cloud-based, the Linux-based ZOE smart home assistant works locally for greater privacy. The company says it plans to make ZOE open source.

A similar approach is being used by Silk Labs, whose AI-augmented Sense surveillance camera and automation hub won Kickstarter funding earlier this year. In June, Silk Labs announced that it was cancelling the product to focus on software and OEM sales. Silk Labs has refunded backers, and has open sourced the Linux-based Silk platform behind Sense. Silk offers local control of AI analytics as well as voice, face, and gesture recognition.

Another open source smart hub with voice AI is Mycroft, which was crowdfunded a year ago, but has been delayed to an expected November launch. Mycroft offers an Echo-like hub that runs Snappy Ubuntu Core on a Raspberry Pi.



Mycroft

ZOE, Silk, Mycroft, and others are pushing a related trend toward AI analytics. The idea is to make home automation more autonomous and self-learning, so devices can truly interoperate intelligently without requiring constant attention via a smartphone app. Some of these systems use voice- and facial recognition to recognize who is in a space, and adjust settings accordingly.

Typically, AI is being introduced in the cloud, as with the optional cloud analytics available with the Nest Cam. As products add analytics, the question is whether local-only systems can compete with cloud services such as Alexa, Siri, Google Assistant, or IBM’s Watson IoT platform. In other words, can you have privacy and security, and also enjoy sophisticated, personalized automation?

In the next two articles, we’ll look at how operating systems are changing to adapt to evolving IoT requirements, and then we’ll finish up with a look at the future of IoT, with a special focus on industrial IoT.

 

Read all the articles in this series

 

 
This article is copyright © 2016 Linux.com and was originally published here. It has been reproduced by this site with the permission of its owner. Please visit Linux.com for up-to-date news and articles about Linux and open source.
 

(advertise here)


PLEASE COMMENT BELOW

2 responses to “Linux-based smart home hubs advance into AI”

  1. chip says:

    “In other words, can you have privacy and security, and also enjoy sophisticated, personalized automation?”

    Short answer: no. It won’t ever happen, at least not in the world as it works today.

    Look at how cellphones are used by govt agencies and think the same thing on a very large scale where a single device can capture many voices in a party/meeting etc. saying something very inconvenient so that access to this intel becomes a valuable asset that can be sold or used against someone.

    By the way: not connected to the cloud doesn’t mean secure privacy-wise, in fact doesn’t mean anything at all.
    If I know you’re home at 22:00 with someone to talk about sensitive stuff, I instruct the device to record everything compressing the audio on the fly on local storage, then the following day the device will ask for a critical update because temperature sensors can’t be read reliably and this would cost money in wrong heating bills or even false fire alarms. You of course allow the update and you bet that the device will spend some time uploading your conversation while telling you that updating drivers takes so much because the network is congested.

    I would be very very careful before putting any of these devices into my home or office, clouded or not.

  2. Tom says:

    As for privacy & security… I sandbox all of my WiFi devices making it difficult to access any open ports in the embedded linux o/s. For remote access there are no open forward ports on the web/internet facing “edge”. All devices are only available via cloud api’s, known trusted i addresses, including video.

    Codechaser

Please comment here...