We know how important EV is detection to our users, especially with our Community. For our first deep-dive of 2020, we sat down with two of the Sense Data Scientists working on EV Detection to get a better idea of some of the challenges and successes we’ve had throughout the years and share where we are now. We hope you enjoy, definitely looking forward to releasing more content like this moving forward!
Cool! But, you do realize there is also a ‘pattern’ (pun intended) in your posts and support answers form your staff? The pattern is always: you do not know, it is hard, ohhh man this stuff is so complicated… we want to - but we can’t … and so on.
There is no difference in this EV post than the other posts about your device detection issues. By now everyone should get it, you can not detect devices reliably and reproducible. So why not let the users know that this is a tough road and Device detection is in fact not a feature right now anyone shoudl rely on. It is a nice to have feature that might be ready in 10 to 20 years time.
Putting out the same arguments, the same explanations is not helping. I do not see the reason why to make the same post for an EV, LED lights, heat pumps HVAC, Furnace or whatever. They are all devices people expect to work and they are all not detected for the same reason…
Would be better to have more of a status report update that says hey we got the system now to detect X-number of devices of TYPE X in 50000 homes. Give us an update how many devices work per week - that woudl be a positive spin on Sense and not what you can not do…
just my 2 cents
Don’t announce features until they’re released. Otherwise you’re just jerking your customers around.
Agree with e.braun
I read this post and then look back to sense on my iPad. Yep, there’s a 11kW load that’s not identified. While I appreciate that Tesla updates firmware and that changes the signature, how many other 11kW loads do most homeowners have? This seems like an easy one to figure out, especially since I’ve told the app I have a Model 3.
If the AI algorithms aren’t perfect (yes, I get the complexity) let the users identify loads. Have a setting for ‘learn’ that I can tell sense what I am going to cycle power. Then with prompts, turn it on and off 10 times (or whatever). Even if this is imperfect, it’ll give the AI a head start and additional data to work with,
I bought the sense for 2 reasons:
1-pre Tesla, see what capacity is left in the panel. For this, sense works perfect.
2-measure actual load/cost of Tesla over time to compare to previous car. For this, fail.