Two collars, a new case and the first detected bark
Head-to-head trips with two collars, a collar that knows when the chase is on, the first detected bark — and a coverage node in the woods.
Since the last post we have been in the woods almost every day. Briefly, where things stand.
Laikku has been wearing two collars, and the same loop has been walked over and over. We are trying different builds, antennas and cases side by side, to find the recipe that carries far and survives the Finnish woods, the rain and the way a dog moves. A few things have already dropped out, and the case has a new, sealed version after that rainy trip.
The collar now knows what it is doing. It tells apart lying in a drawer, a sleeping dog, a walk and a chase, and when the chase is on it raises its update rate on its own. This was tuned on real recordings from real dogs: a sleeping, scratching, garden-roaming or car-riding dog must not raise the rate, because then the battery would drain at a walk.
And for the first time the collar recognised a bark. First from the recordings: it found barking on exactly the trips where I had noted barking, and nowhere else, not shaking, not the car, not sleep. And then for real on today's trip: Laikku barked twice, the collar reported both, and never once for nothing. No sound is recorded or sent anywhere; the collar decides from motion alone.
Coverage nodes have gone out into the terrain. On the latest trip one was strapped to a tree mid-walk, where the dog usually works. A node does two things. It stretches the reach of the local radio: a collar that would already be out of earshot reaches the phone through the node. And once the node reaches a cloud node, there is no limit left: the dog shows on the map at home and on a friend's phone at the other end of the country. On this trip whoever sat at home saw the dog for the whole walk through that one node.
And a pile of things that show nowhere but are done.