I would appreciate some direction on where to start to fix an inconsistent behavior - which I strongly suspect is caused by me
Hardware: _Intel 5i/ 8GB /500 GB
OS: _Windows 7 running OH3 and MQTT broker
Java Runtime Environment: OPENJDK 11
openHAB version: 3.0.2
I have connected my PowerSeries DSC Alarm panel to OH3 through use of an ESP32 module as described here . Works like a charm.
My setup is Google assistant enabled. Thus, when I say “OK Google, switch off the Away Mode” and if the alarm is armed, then GA says: “OK, disarming the Alarm”. If the Alarm is already disarmed GA responds “The Alarm is already disarmed”
Makes perfect sense.
The inconsistency comes in with the gate.
I have a two limit switches which senses if the gate is fully open or fully closed. This is linked to a ESP32 which talks to MQTT which talks to OH3 - standard design for these type of equipment.
When the gate is fully closed and I ask GA
“OK Google is the gate open”,
GA responds “The Gate is closed”
This is correct -since the system knows the gate is fully closed due to the limit switch.
However, if the gate is fully closed and I say: “Ok Google, close the gate” then
GA responds in saying: “Sure, closing the gate” but this incorrect since the gate is already closed.
I would like to have a similar behaviour between gate and alarm. If the alarm is disarmed and I give the disarm instruction I am told it is already disarmed. Likewise, if I instruct the closed gate to close, I want to be told that it is already closed.
Not sure how to share my gate and alarm setup in OH3 with you guys, but any pointers on where I can start investigating would be much appreciated. If this is documented and I have missed it, happy to get a pointer to an URL as well,
This is just how Google Assistant works. If you ask it a question, it answers. If you give it a command, it says that it’ll carry out the command and then does it. So it’s not incorrect, because Google is actually sending a command to close the gate, and you should see it in your log as a received command. OH also isn’t smart enough to reject the command. It just passes it on, and nothing happens since the gate’s already closed.
Your command to close the gate actually has your first question (“is the gate open?”) built into it. Humans instinctively process this IF statement and respond that your command doesn’t make sense since the gate is already closed. Google Assistant doesn’t yet have that logic, so it needs you to ask if the gate is open and then command it to close the gate if necessary.
I imagine we might see this down the line, as there are many ways in which the Assistant AI could be improved.
Then we’ll eventually get people asking the same question that you are, because it will still be inconsistent. Lots of people use fans, blinds and TVs, and they might feel differently about what would be included in “most scenarios”.
Ultimately it’s up to Michael to decide how much time to put into it and how “complete” it needs to be. I don’t have the programming skills to help, but if you do then I’m sure it’d be appreciated.