1NF053C
commented
1 year ago Strongmanning: An argument for not accepting this PR into main is that if at least 50% of maps start an HQ that has a long enough obstacle between HQ and the first sensed well (and concaved towards HQ), then it would consistently build carriers one step in the wrong direction.
I'm not convinced that at least 50% of maps would have this characteristic, although this isn't something I can be certain about, so I'm open to being convinced otherwise or even more subjective preferences here if there isn't map data we can objectively rely on.
Future: I think HQ build direction is a smaller problem within a larger problem of pathfinding. I think the way to improve beyond a Greedy Best First search approach in general would be to change BugNav navigation into Dijkstra or A* Navigation, if we ever find the time to do this.
Dependency on Comms: We'd probably would want to have every tile of the map serialized in Comms if we want to implement more advanced pathfinding.
Micro-enhancement that I'm using during testing carrier conveyor built implementation. This feature has same trade-offs as a Greedy Best First Search strategy. (Our BugNav is itself a Greedy Best First in that it too is tricked by obstacles up until the moment of collision).