machine learning requires “ground truth” which for purposes of Sense’s application is relative good time resolution power usage info. It’s not like a “library” of smart devices is going to give them what they need.
Sense has done a pretty good job integrating with devices that give that kind of “ground truth” feedback - smartplugs plus Philips Hue (which does provide power information, not just on/off).
The one place where Sense is using on/off network data, NDI, seems to work OK for some smart TVs, but relies on low latency time between the device waking up on network and the associated power transition on the Power Meter.
Not sure an inventory of smart devices plus relatively high latency info about whether they are on or off is especially helpful.
Probably true, but at the end if the day I’m interested in having as much of my energy usage reported AND identified as possible . I’d like to be able to use my Sense app to see a monthly report of how much things like my various led recessed lights I have installed different rooms, including my outside lights, and are connected to SmartThings energy reporting dimmer switches costs me.
Have to agree that linking to ST would be useful. I understand that it will not help with the machine learning magic. But let’s be honest, that mostly doesn’t work any way. Most of my usage is always on or other. Devices that have been identified suddenly stop using power, and then a totally new device is discovered.
At least if there was ST integration they could point to a jump in usage and then indicate “ST said this device was turned on within x seconds of this jump in usage”. 100% accurate? Surely not, but could probably be better than what we currently get.
To make Sense detect devices in my house which it did not in a year, I resorted to using TP link smart plugs to all m y major applicances. This is great because now I got my always on bubble in sense to go down. Next things is for smarthings to have the data about each device’s usage propagated so that I can write some pistons based on energy usage of each appliance. Is this possible as of today with the unofficial handlers? I tried SmartThings_SenseMonitor by brbeaird but it does not break down the individual devices for me
I am with you. If Sense integrated with SmartThings, I could shout my deep well pump off if sense showed it on too long indicating a leak somewhere on the farm
I’ve had my Sense and Smartthings systems (Comprised of arlo, Luton, ge, ecobee and a random mixed bag of sensors and relays) working independently for 4ish years and I actually have a real need to integrate the two. I can’t believe this isn’t officially supported yet. This seems like an obvious upgrade to sense for greater exposure through the Smartthings ecosystem.
I need to load shed if certain loads come on in the house. So the “linked services” should share total load and when certain loads come ON or OFF to smartthings. I’d also like to see solar data being shared to smartthings for when I finally pull the trigger on that project.
Thanks… and kindly hurry up with the official support .
Respectfully, I think you’re barking up the wrong tree.
Sense has an API and cloud integration capabilities for this already and therefore they have done their part on the “official support” side. The SmartThings community released a Device Handler for this 2 years ago GitHub - brbeaird/SmartThings_SenseMonitor: Connects SmartThings with Sense. While the community device handler is meant for the old app, that’s not on Sense to resolve. I would suggest you talk to brbeaird and ask him to update the existing device handler.
As a third option you could used Home Assistant (which is free and plays well with SmartThings) Sense - Home Assistant. With HA you could do everything you’re asking for. HA relies on the cloud API to get all the data necessary for your listed items.
The majority of this thread is actually the opposite of what you’re asking for. Most people, myself included would like to see a way to feed data to Sense in order to help Sense recognize more. Although I use Hubitat instead of ST, I have over 100 devices that monitor electricity usage including every outlet and light switch in the house. That IS something that Sense would have to allow us to do and why many in this thread are asking for Sense to step up.
Maybe you should consider asking Brian (who made the DH) to hurry up with the support
Thanks Todd for the thorough and detailed post. Much appreciated. I did find the git hub community device handler but saw its rev date and quickly dismissed it – coming straight here to complain on the 2018 post.
I’ll explore some of the avenues you recommended for a quick resolution to my load shedding scheme, but still believe that an official “connected app” vs API would be beneficial to sell more sense devices. It’s actually why I chose the cameras (w/hub), thermostat and lighting control system(w/hub) im currently using - the integration was so easy and all in 1 app (which I both love and hate).
Is there any progress for SmartThings integration yet? I’m new to Sense and have noticed around every turn that they want data to help with their machine learning algorithms, such as; access to my network for Network Device Identification. How better to help identify devices than through my ST Hub.
Quite honestly, I waited several years before purchasing Sense because I wanted to wait until it was supported by ST. The only reason I made the jump now is because I updated my electrical system and added a sub-panel for my new portable generator. The point is, how many other people like me are waiting for ST integration before taking the Sense plunge?
I think Todd gave a pretty good assessment up above. The user community has developed a turnkey integration for Home Assistant and there is an integration path for Hubitat. I don’t think Sense has really focused on hub-based integrations, in favor of power monitoring device focused integration, for a few reasons:
Much of the device data from hubs is simple on / off or dimming data - far less useful for learning.
Hubs often have data collection latencies due to the slower communication protocols (ZWave, Zigbee). Sense is based on near real time data collection.
The Hue hub actually covers both of these so it was a natural - provides power data and low data latency.
I understand the desire for a SmartThings integration, but there are constraints on what type data is actually useful to Sense for machine learning, vs what a user might see as useful.
FYI - I actually roll all my energy usage up to Home Assistant because it has Sense, Tesla, Ecobee, electric/gas meter, and Lutron integrations (Sense bundles the Hue and TP-Link smartplug data so I don’t need to use those).
The explanation you offered, thank you for that by the way, would be a direct contradiction to what Sense has claimed to be their business model. Data from a wide array of inputs, especially one as popular as SmartThings, contains hoards of information that can be leveraged for more consistent device recognition.
The contradiction is that Sense is already asking for NDI (Network Device Identification) permission to help with their algorithms. This would have the same latencies as SmartThings. Further contradiction is the fact Sense has similar integrations in use, such as; Alexa, Ecobee, etc.
The simple on / off or dimming data referred to in the 1st bullet point is exactly what would make this integration a powerful data source. The ultimate goal is to help identify devices through so called machine learning using a variety of data inputs. Therefore, simple on / off or dimming data can help identify and/or verify devices.
One of the more prevalent needs SmartThings Integration can help with is disseminating between devices with near identical waveform signatures. If I ask Smartthings to turn on my stairwell light and it has a near identical waveform as my nearby reading lamp, Sense will have that collaborative data to know which device it is without having to dump it into the “Other” bubble.
We don’t need to reinvent the wheel here, just trying to use the tools available to help with device identity. What better place than a very popular platform that many people already have. SmartThings integration does not even need to come with any added bells and whistles, monitoring and reporting triggers to Sense and perhaps the occasional device identification verification by the user would suffice.
@kevin1: Upon reading your reply it makes me wonder if you personally do not like SmartThings or am I reading more into it (which I definitely have a habit of doing) than there is? Please forgive me on the topic if I am way off base, you are obviously much more Tech Savvy than myself. I’m still stuck on the fact that Sense “appears” to already use identical reasoning with their integrations of; Alexa, Ecobee, etc., and justification for asking permission to access my network for NDI.
@LarryME,
All good thoughts. But let’s push down a level on each of these.
Generic on / off data, especially lagged and possibly packetized, isn’t good enough for current mostly-unsupervised machine learning at Sense AFAIK. Maybe in the future if Sense starts doing more supervised learning. And quite honestly, I don’t believe NDI has worked all that well exactly for that reason.
Your Alexa and Ecobee examples aren’t really valid. Alexa’s a whole different bird, not related to machine learning. And the Ecobee data is only used on the back-end for very specific model development, HVAC, not direct machine learning, because it is only every 5 min sampling. That’s why the integration is called Ecobee Historic - because it isn’t realtime. I guess that Sense could pull historic data from SmartThings as well, but the HVAC was targeted because it is such a big, and yet tricky to detect, consumer of energy (especially heat pumps and mini-splits). I can understand why Sense, that had a choice of a grabbag of data from a hub, vs highly targeted data for one specific set of model types, I would pick the latter.
I don’t have anything against SmartThings. Just pointing out what works today and what doesn’t, and the thing that has worked is hub developers that are willing to integrate to Sense using existing hooks. There are also some technical constraints that limit SmartThings, given the current Sense capabilities - communications protocol used by many of SmartThings IoT device (Zwave, ZigBee) and the total mixed bag of incoming data.
Not trying to be an apologist for Sense, but rather inject a little bit more logic and technical info into why some integrations might be harder or less useful than others.
Straight on/off information with some latency may not be useful, but on/off information on a slight delay could still be useful if it also included power consumption (but probably not directly useful for device detection). This is, as I understand it, what you get with the existing smart plug integrations. Sense doesn’t get real-time data from the plug of a power change; it gets the real-time power change from the direct meter measurement and a (usually within a couple seconds) power reading from the plug.
Given that, I don’t think an integration with ST would be useful beyond providing an on/off toggle like what (used to?) exists for the smart plugs. Consider the problem facing Sense without only on/off data: that jump in usage that happened within a few seconds of you turning on your kitchen light…was that all just the kitchen light, or did your computer’s virus scanner just start during that same “few seconds” window? Sure, I guess, given enough cycles, you could start to learn what “turning on the kitchen light” looks like electrically, but, whoops, a bulb just burned out, and I didn’t notice it, or I did, but I’m busy, and it’s still light enough in the kitchen that it’s not a high priority to replace it. Now I’m adding garbage data to the mix, and if I don’t replace it with an identical bulb, the new reading could be off by a couple watts one way or the other.
I do own a ST hub, and I would love it if those problems could magically be solved, but I don’t see that happening anytime soon, and I’m OK with that.
I seriously don’t see why Sense developers and engineers are not building this integration! This thread is probably one of the biggest but nada from them.
@baivab.mitra
If I look at the wishlist sorted by number of likes, there are several other more “requested” enhancements than this one, including a HomeKit integration. Please note that Sense uses top level “Likes” when doing their assessment so this wishlist item is at 60 “likes”, not the 109 in the list ordering here, and the other leaders need to be suitably adjusted as well.
Beyond the end-users demand issue, there’s also the question of whether the integration would be possible to implement and whether it would have any additional value to Sense. Integrations to smart plugs are useful because they provide full “ground truth”, power data from individual devices. My thoughts (not Sense’s) in the link below.
Finally there’s the question of who should do the integration. There’s already a great integration with Home Assistant, because users of both Sense and Home Assistant were willing to develop a very robust turnkey integration built on Sense’s informal API. Maybe an enterprising SmartThings user would be willing to do the same ?
I’m curious as to what you would want the integration to do? This thread has many different asks for a ST integration and the OP appears to want device control.
I’m more in the camp where I would like to see Sense stick to energy monitoring and completely remove the power on/off capabilities from the Sense App. I think that it gives a false impression that they are a home hub of sorts. Since all of the devices with power control capabilities are via 3rd party, there’s other ways to do that. Additionally, Sense doesn’t have tiles, or easy access to these power control features which are common on hub UIs.
.