Must be reasonably high powered hardware along with unbloated OS (Linux)…
Local Operation is available now: just not from SmartThings
There are multiple home automation systems available in the low cost range which are primarily local. Some of these, such as homeseer, are at least as powerful as SmartThings. Many preexisted smartthings, some are new. (Apple’s HomeKit, for example, runs everything locally except voice recognition, and that can be done over cellular as well as over Wi-Fi.)
If you want local operation, there are lots of competitors to choose from.
SmartThings and the Leapfrog game
What’s SmartThings initially did that was different, and what gave it the big press buzz in 2015, was offering a combination zigbee and Z wave system in a low cost range which allowed customers to do two things: write very complex logic for their rule sets which would be hosted by SmartThings (smart apps) and write their own custom device handlers. All in a system where the mobile app screenshots looked consumer-friendly.
And their cloud-based platform enabled easy integration with a lot of the newest stuff that was coming on the market. So they had a Phillips hue bridge integration, an IFTTT channel, Amazon echo integration, and unofficial nest integration long before their pre-existing locally-based competitors.
All of which made smartthings the “ooh, shiny!” Home automation story for 2015.
But it’s 2018 now. The competition has caught up. Multiple companies now offer both zigbee and Z wave. Many have added cloud integrations for at least the most popular devices. It’s true that a lot of them still don’t let you write your own device handlers except maybe as plug-ins, but that’s becoming less important.
SmartThings’ next big jump is coming
People have been talking about the benefits of local processing for Samsung SmartThings since Samsung bought it 4 years ago. Not much has changed with the exception of some limited features and smart lighting, and that option doesn’t run custom code.
Samsung’ new platform is in development should be here within a year. Again, lots of discussion in the forums about it. It probably will have more local processing, and it will definitely be more reliable. But custom code will likely no longer be hosted in the primary cloud. And at the present time individuals are limited to 10 custom smartapps, although that will probably change.
It looks likely that long term you’ll still be able to write pretty much anything you want, but with a big difference: you’ll have to host the custom pieces yourself. That’s good for overall stability, but does make things technically more complicated for the individual developer.
The hardware Samsung loves to sell
Samsung is primarily a hardware company, with a ton of different lines and lots of patents. (No patents for the SmartThings hub, however. Which tells you a lot if you pay attention to that kind of stuff.)
And do you know what they really like to sell? Televisions.
Based on these trends, Samsung continues to strengthen its leadership in the global TV market. According to GfK and NPD, between January and August 2017, Samsung was the number one player, recording 34 percent of total global TV market revenue, higher than the market shares of the second and third largest players combined. The company also recorded 42 percent for big screen TVs, 60 inch and larger, and 38 percent for UHD TVs. Samsung also led in market share based on TV pricing, with 44 percent in TVs, priced over USD 1500, and 37 percent for TVs over USD 2500.
Also high-end appliances and cameras.
They want to seem high tech, easy to use, sexy, and worth very high price tags.
Samsung sells an expensive (over $3,000) unit in this category every second in the United States.
Those are the hardware sales they care about. That’s the market segment they care about. Plus, of course, some mobile phones. 
Their vision of IOT fits into their vision of appliance and television sales. It’s an add-on feature. They don’t really care whether you use it or not as long as the feature being available makes you buy more of their appliances and televisions.
So…none of their development decisions are likely to be based on what people in this forum have been asking for four years. Or on how many $99 hubs they can sell. Or on the logic of local processing versus cloud for IOT. Over on how “simple” things would be with a different paradigm.
They already have a new paradigm coming.
As for the rest: if you want an IOT system for your own home based on local processing, there are a lot of alternatives out there you can consider. But you’re not going to change the course for this ship now.
Just sayin’… 
2 Likes
The speaker is near the microphone, it’s not natural language processing but rather a quite limited set of specific 2 or 3 word phrases, and it’s a pretty complex processor.
One of the problems you run into with home automation is the number of different devices and the number of different names that they have. This can rapidly crater a simple voice recognition system.
That is, if it just had to recognize “lamp,” “lock,” and “Plug” you could probably do it locally. But as soon as we start doing things like “Michael’s lamp,” “kitchen light,” and “coffee maker” we have vastly expanded the vocabulary of what it has to recognize. So basically once you let people name their own devices, you can’t use the “simplified voice commands” that the car systems offer.
While the list of commands in my Honda is limited, the amount of street names, cities, song and artist names that the software has to recognize is quite lengthy. I’ve never researched what type, power and cost the computer hardware in modern vehicles have. Of course they have scale on their side to keep costs down. But I would have happily spent several hundred on a ST-like Hub that performed more reliably.
As far as whatever new apps they come out with, I sure hope a little more thought goes into the naming of it:scream_cat:. I have always like version numbers - like ST v1, ST V2… Even newbies could guess the chronology 
1 Like
The street names and song names are all from a pre-populated database. It’s not letting people name things for themselves. 
1 Like
SmartThings now has the same “experience” (including the advantage of all the code and exact usage data for optimization purposes) but they choose to set other priorities.
Power reporting (Watts) from ZigBee outlets makes up over 80% of Event traffic to the SmartThings Cloud (per ActionTiles analysis). That’s just super wasteful and trivial to fix (ZigBee reporting intervals on the Outlet are configurable, but hard coded in the DTH).
Regardless, with the resources of Samsung, SmartThings could have dedicated a team to local processing and had it solved several times by now.
Samsung’s priority is the merger of their multiple IoT clouds into the SmartThings Cloud and App.
-
It’s the foundation for everything, including any future Hubs that execute more stuff locally.
-
Smart Lighting is already local.
-
Average SmartThings customer has a handful of Things (well under 20) and does not have many critical automations. Forum members are atypical by far. The IDE was intended only for developers.
It’s not intended as a slur, but SmartThings is, intentionally or just de facto, using us as an extended Beta test.
And given that their cloud and support is “free for life”, and customers expect continuous improvement, ipso facto, the current version is always imperfect and incomplete.
This product comes out of a completely different culture than anything I’ve ever experienced. Samsung may try to shift the culture; especially to coincide with the latest radical changes (the new App is a HUGE change which I think we’re essentially ignoring!).
The most positive cultural change for this Community would be if Developer Advocates would be re-added to the staff.
The biggest change to other consumers takes longer: Globalization, faster growth, and progress on “average customer” pain points, especially as these households become more sophisticated and have more Things. Migration Tool? - inevitable, but it still might take 5 years (approx 5% of all households reset or replace their Hub or Location at least once after purchase).
2 Likes
But the software still has to translate anyone’s voice to those tens of thousands of city, street, song and artist names. Very much like “Michael’s kitchen “ or any other device name one might conjure up…
It would be nice if ST staff perused these forums and interacted with their customers. But a lot of companies just don’t value customer feedback. I don’t know why🤔. It can’t be the cost of paying one or two staff. Just corporate apathy? I was monitoring a post about an email bug in the Apple forums that affected 10,000+ users, but Apple never said a word… I find that lack of involvement to be unprofessional. Maybe I just expect too much out of my vendors🤔
1 Like
Very different problem in terms of voice recognition. There are a lot of technical papers and presentations on this type of thing if you are interested. But essentially matching to an already known set of syllables is very different than parsing natural language.
It would require a systen built with the same sort of framing that the car systems use:
Start
Human: “Automation”
System: “What do you want to do?”
Human: (From a selection of predefined phrases) “Turn on”
System: “Which device?”
Human: predefined Device name
And as shown in the Audi video, the car systems do get even the predefined phrases wrong fairly often.
I can’t answer for “a lot of companies”, but here’s my perspective:
-
I know that SmartThings staff at all levels of the organization read the Forums … some more regularly and thoroughly than others. Yes - we only see rare responses from a few people, but there’s plenty of good reasons for staff to have “no comment”.
-
I run the ActionTiles “Feedback Forum”, and find the input very valuable, but it’s very difficult to write “official responses”.
I think companies do very much value feedback. What they don’t relish, however, is having to respond to that feedback with anything more than a generic prepared statement; i.e., “Thanks for your feedback. It’s a great idea that we’ll pass back to our development and product department!”.
You see, by saying anything else than that, the company is making a promise of some sort. Either a promise to add a feature or fix a bug - and if no timeframe is given, you can bet the next “feedback” is a request for “when?!?” - or an indication that the feature or bug is not considered a high priority, thus instantly disappointing the customer, making them feel as if their feedback is not appreciated; and generally worse than if they had not been given the opportunity to give feedback at all.
Responding to feedback with a feature promise or roadmap also gives competitors a strategic advantage; as they can use this information to accelerate their own competitive features.
I don’t know about you, but I’ve been on a hundred “customer service calls” with dozens and dozens of companies, a large proportion of which offer a “satisfaction survey” at the end of the call (or by an email survey follow-up). The surveys sure feel like BS. I’ve given both good and “highly-dissatisfied” scores; but in neither case did the company respond (except, as mentioned, sometimes responding with a form letter).
Companies want feedback. But there’s really no net value (at least relative to risk) in attempting to complete the feedback loop. 
FWIW, I have had both Apple and Amazon respond individually to my feedback comments on accessibility issues on several occasions, including requesting more details and escalating the issue to technical staff. Both companies seem genuinely committed to “customer experience” quality, including some kinds of edge cases. These were private discussions, not public forums, however, which goes to your point.
No promises were made, but at the same time I felt they were sincerely interested in customer experience details.
One Apple rep called me by phone once a week for 3 weeks until an issue with my Apple Watch had been resolved (which I had reported as a general feature request, not a warranty problem).
Amazon has also been very interested in details and in collecting information on what I felt would work in my situation.
In both cases features were added about 3 months later which significantly improved accessibility. Could be just a coincidence as I’m sure I’m not the only one who had the issues, but both companies have demonstrated to me the quality of active listening, and both have offered steady improvements to accessibility over time, which I very much appreciate.
So I do think different corporate cultures approach these issues differently, FWIW.
2 Likes