Real-Time Path Replanning
Without Recalculation
In real-time systems, paths don't stay valid. Targets move. Environments change. Agents adapt. Traditional pathfinding breaks here — because every change requires recomputation.
The Problem with Replanning
With A* and traditional systems: goal changes trigger recomputation, environment changes trigger recomputation, blocked agents trigger recomputation. Each recomputation is as expensive as the original path search. In dynamic systems, this happens every tick — creating CPU spikes, unstable frame times, and cascading performance issues.
Why Replanning Doesn't Scale
Replanning assumes each path must be recalculated from scratch and each change invalidates all previous work. In a system with 100 agents that retarget once per second, that is 100 full A* searches per second just for replanning — on top of the initial pathfinding cost.
The StrataNav Approach
StrataNav separates navigation into two layers: precompute (once) and query (infinitely). When goals change, the navigation field does not change. Agents simply query new directions from the existing field. Replanning becomes a lookup, not a computation.
Performance
StrataNav delivers 13× faster replanning in query-only mode. Change goals every tick. Update agent direction instantly. No path rebuild required. No CPU spikes. Stable frame times under dynamic conditions.
Use Cases
Real-time replanning is critical for: RTS games (units retarget constantly), MOBAs (dynamic engagements), action games (AI reacting to players), crowd simulation (continuous flow), and robotics simulation (dynamic environments).
The Architectural Shift
Most teams try to optimize replanning. The real solution is to eliminate replanning cost entirely. When navigation is infrastructure rather than computation, goal changes are free.