Each delivery scored using f(n) = g(n) + h(n). g = actual travel cost with traffic weights, h = heuristic distance to destination. O(n log n) vs greedy O(n²).

Nearby orders grouped using nearest-neighbor clustering before vehicle assignment — reducing total distance by batching pickups in same zone.

Every 2 seconds, new orders trigger a re-prioritization pass. Vehicles reassigned using priority queue sorted by urgency × distance × capacity.