The 3D pathfinding & volumetric navigation plugin for Unreal Engine 5
Sparse Voxel OctreeGPU voxelization10,000+ agentsMultiplayer replicationBlueprint & C++
LCM 3D Navigation AI brings true 3D pathfinding to Unreal Engine 5 — flying AI pathfinding for drones, dragons, spaceships, helicopters and underwater agents that Recast navmesh simply cannot express. Unreal's Recast navmesh answers one question: where can I stand? It flattens your level into walkable polygons — exactly right for a soldier, completely wrong for a dragon. LCM 3D Navigation AI answers a different one: where is there empty air? It carves the whole playable volume into a Sparse Voxel Octree and paths through the empty nodes, so open sky costs almost nothing and detail appears only where geometry actually is.
01Sparse Voxel OctreeThe level's empty air is subdivided only where geometry exists — one node covers open sky, so cost tracks clutter, not level size. Drop a single LcmNavigationManagerSVO per level and it owns the octree, the pathfinder and every query.
02Three solvers, three smoothing modesA* for raw throughput, Theta* for true any-angle routes, Lazy Theta* for near-Theta* quality at near-A* cost. Then Raw, Linear Shortcut or Curved Spline smoothing — validated against geometry, so a smoothed path never cuts a wall.
03Macro / micro routing & infinite worldsLong routes solve coarsely across chunks, then finely within them. Finite worlds build once at BeginPlay; infinite worlds stream chunks around agents and throw them away behind.
0410,000+ agents through Mass ECSTwo crowd modes: per-agent asynchronous pathfinding via the Nav3D Agent trait, or one shared CPU flow field sampled O(1) by an entire swarm. Three auto-registered processors run request → solve → steer; you never call the solver. A per-agent GPU solver is in experimental validation — coming as a free update.
05Tactical EQS in true 3DGenerators and tests that score real volume — volumetric line-of-sight, cover, threat exposure, height advantage and true path distance — so a sniper perch is chosen in 3D, not on a floor plane.
066-DOF flight & swim locomotionNavigation gives guidance; LcmFlightMovementComponent turns it into a body. Multirotor, Bird, Fish and Generic 6-DOF archetypes with thrust, lift, drag, buoyancy and attitude PID — plus an animation-drive struct for blendspaces.
07Server-authoritative multiplayer replicationClients request a route, the server solves it and delivers it back through Client_DeliverPath; moving obstacles replay their nav footprint to every peer over a multicast. A headless validation suite (-LcmMpTest) exercises the sanity, replicated-request, multicast-transport and digest-chain paths, and CPU compute stays bit-deterministic for replays and lockstep.
08Built for real projectsDynamic obstacle restamping, AI Perception threat tracking, hybrid Recast + SVO levels, nav links, area modifiers and a live Gameplay Debugger.
Whichever way your team authors behaviour, LCM 3D Navigation AI plugs into it — the same navigation data, three authoring surfaces.
Behaviour Tree
Drop-in tasks, decorators and services — FlyTo, line-of-sight and chunk-loaded checks — alongside seven documented stock-node patterns showing where an LCM node earns its place.
StateTree
The same flight behaviour authored as StateTree tasks, so state-driven agents get identical routing without a Behaviour Tree in the stack.
Mass Entity
Two ECS traits — per-agent pathfinding or a shared flow field — with three auto-registered processors driving request, solve and steering for 10,000+ entities.
More pre-built nodes and authoring shortcuts are on the way — shipped as free updates.
Platform support
Shipping-ready on PC and current-gen consoles, with the C++ core built to the same standard on every target.
Windows PCSupported
PlayStation 5Supported
Xbox Series X|SSupported
Nintendo SwitchComing soon
Next · v3.0
In development
Free update for existing licences
GPU path solving · experimental
A per-agent GPU path solver and GPU flow field — currently in experimental validation through our research programme — plus further profiling to widen frame-budget headroom at crowd scale.
Replication improvements
A wider server-authoritative surface and lower-bandwidth path and obstacle sync for larger multiplayer sessions.
Mixed pre-baked + realtime infinite worlds
Bake the authored parts of an open world ahead of time and stream the rest live, in one navigation graph.
More pre-built AI nodes
Additional Behaviour Tree, StateTree and Mass authoring nodes so common jobs need less wiring.
Documented like a product, not a plugin dump
A 128-page manual across 17 chapters, free to read online — every reflected symbol carries a tooltip, and every setting is listed with its verified default and limits.
Production-grade Niagara built the same way as the plugins — profiled, documented and shipped in real games. Explosions, energy beams, muzzle flashes, shields, portals and full environmental systems.
L. Charitha MadhushankaSenior Game Developer · Unreal & Unity
UEDEV is the Unreal engineering label of L. Charitha Madhushanka — a senior gameplay engineer shipping commercial games, VR/AR, high-performance C++ plugins and engine tools. For studio work, contract jobs, or a bespoke plugin or system, head to the full developer portfolio.