I’m having trouble wrapping my mind around how I might achieve this concept. I’ll start with a bit of background on what I’m already doing and where I’d like to get.
I currently use multiple zones, coupled with the native Person component and multiple tracking technologies for each person, such as BLE, Life360, Mobile App, etc. I use these zones to announce using TTS when someone has entered or left specific zones. It works reasonably well. But I believe a well known weakness in this implementation is how the Person component is designed to simply use the last updated entity, rather than to intelligently look at all the entities assigned to a Person. Therefor if a person is sitting at the edge of a zone for example, TTS will announce their departure and entry to the zone as the various inaccurate GPS readings kind of jump around far enough to cause a change, even though say 3 of 4 ‘sensors’ say the person hasn’t moved, the 4th sensor wins anyway.
Enter the Bayesian sensor. Unfortunately, it looks like the only way I can use the Bayesian sensor in this concept is to have a input_boolean for each zone and automated around that. I’d really rather not create a bool for each zone. Seems like person tracking is un-needlingly complex in HA because some core concepts are not implemented in the Person component.
Does anyone have an concepts or examples they can provide beyond what I’ve conceptualized, that may simplify this a bit in setup?
Hard to believe someone hasn’t encountered this at this point. The in-built person solution just doesn’t work well with multiple device_trackers per person.
Not room level. I’m talking GPS type positioning with several zones around town. Each person has at minimum 3 types of location systems. BLE and x2 GPS. The Person Zone detection goes with the last updated device_tracker. So it’s possible that when two or more don’t agree to whether someone has entered a zone that they can be reported in the zone and then immediately reported back out. Unfortunately GPS accuracy is inconsistent.
I already have a Bayesian sensor for home, but I’d rather have a solution that is one sensor that is capable of multiple Zone reporting. So if, for example, my kid is at home or at school, the single sensor reflects either one.
Right now, the only way I can think to accomplish this is with multiple Bayesian sensors and then automations to aggregate them. That’s rather messy,