Yeah, I’ve spent a lot of time on that. Start by adding a cost multiplier that accounts for terrain uncertainty – make the algorithm treat a blocked node not as impassable but as high cost. Then run a quick sanity check with a sliding window: if the heuristic over a small area spikes, re‑evaluate the local map instead of waiting for the whole path to recompute. That keeps the flow smooth while still adapting to messes. Keep the code modular so you can swap in a better heuristic later without rewiring everything.