Navigation Bundle 2.0 combines Swarm Core, Flow Fields, and Steering + Avoidance into a scalable movement stack for crowds, RTS units, and flying swarms.Swarm AI Toolkit - Navigation BundleThe Navigation Bundle combines the main movement systems of the Swarm AI Toolkit:Swarm AI Toolkit - CoreSwarm AI Toolkit - Flow FieldsSwarm AI Toolkit - Steering + AvoidanceTogether, they provide a scalable, system-driven solution for simulating, navigating, and locally steering large groups of agents in Unity.This bundle is designed for projects that need stable large-scale unit movement without relying on Rigidbody physics, heavy per-agent pathfinding, or NavMesh-driven crowds. It supports flat XZ movement, terrain-following agents, and flying-style 3D swarm setups.What This Bundle ProvidesSwarm AI Toolkit - CoreA high-performance agent simulation foundation featuring:Centralized system-driven updatesLightweight agent movement2D and 3D spatial hash queriesPhysics-free movementJob-system movement groundworkSeparation supportOptional debug visualizationCore is responsible for efficient agent simulation and neighbor lookup at scale.Swarm AI Toolkit - Flow FieldsA grid-based global navigation system that:Generates shared direction fieldsAvoids per-agent pathfindingSupports single-layer and layered navigationSupports dynamic targets and dynamic costsSupports terrain, raycast, fixed-altitude, and heightmap-based height workflowsIntegrates with Swarm Core movementFlow Fields provides global navigation direction for large groups of agents simultaneously.Swarm AI Toolkit - Steering + AvoidanceA local movement layer that provides:Predictive local avoidanceORCA 2D reciprocal avoidanceORCA 3D-style avoidance for flying agentsContinuum crowd avoidanceDynamic obstacle avoidanceSteering behaviours such as seek, flee, arrive, follow, pursue, wander, formation, boids, separation, cohesion, and alignmentAgent priorities, weights, smoothing, and debug visualizationSteering + Avoidance handles local collision resolution and natural movement on top of global flow direction.Designed ForRTS unit movementTower defense wavesMOBA-style minionsLarge crowd systemsSwarm simulationsStrategy games with mass unit controlTerrain-following ground agentsFlying swarms and layered navigation setupsWhy This BundleBy combining Core, Flow Fields, and Steering + Avoidance, this bundle provides:Efficient large-scale simulationShared global navigation directionLocal avoidance and natural steeringStable performance at scaleClean modular architectureNo required NavMesh dependencyNo required Rigidbody overheadSupport for ground, terrain-following, and flying agentsIt forms the movement foundation of the Swarm AI Toolkit ecosystem and can be used on its own or extended with additional AI modules such as Perception and Decision.Engine & CompatibilityUnity 2021 LTS or newer recommendedCompatible with Built-in Render Pipeline, URP, and HDRPNo render-pipeline-specific runtime dependenciesNo required third-party runtime librariesPC, console, and mobile supported, depending on agent count and settingsArchitecture OverviewThe Navigation Bundle consists of three modular movement systems:1. Swarm Core - Agent Simulation LayerCentralized system-driven update loopLightweight agent movement2D planar and 3D spatial query supportSpatial hash grid for neighbor and proximity queriesPhysics-free movement modelRadius-based agent representationManaged fallback pathBurst/job-system groundwork for movement, spatial hash rebuilds, and separationDesigned for large agent counts2. Flow Fields - Global Navigation LayerGrid-based cost field generationShared direction fields for groups of agentsSingle-layer and layered flow-field navigationMulti-target direction field supportDynamic target updatesStatic and dynamic cost stampingDynamic cost contributorsTerrain, raycast, fixed-altitude, and baked-heightmap height workflowsNo per-agent pathfindingDesigned for mass movement scenarios3. Steering + Avoidance - Local Movement LayerPredictive agent-agent collision avoidanceORCA 2D reciprocal avoidance optionORCA 3D-style avoidance for flying agentsContinuum-style crowd avoidance optionDynamic obstacle avoidanceCost-field and obstacle-field avoidanceSteering behaviours including seek, flee, arrive, follow, pursue, wander, orbit, density, formation, boids, separation, cohesion, and alignmentVelocity blending, smoothing, weights, and prioritiesAvoidance urgency reporting for stable behavior under collision pressureMovement PipelineGlobal direction from Flow FieldsLocal avoidance adjustmentSteering blend and behavior weightingSwarm Core movement integrationOptional terrain or fixed-altitude projectionFinal agent position and velocityPerformance CharacteristicsDesigned for large-scale simulations with hundreds to thousands of agentsPerformance depends on:Agent countFlow-field grid resolutionNeighbor query radiusMax neighbor countAvoidance module settingsUpdate rate and rebuild intervalsNon-alloc query paths used where possibleSpatial hashing reduces broad-phase query costDynamic flow-field rebuilds can be skipped when contributors have not changedNo Rigidbody components required for default movementNo Unity NavMesh dependency requiredSystem-driven update model with lightweight agent componentsMovement ModelSupports planar XZ movementSupports terrain-following ground agentsSupports fixed-altitude and flying-style agentsSupports 3D neighbor queries for flying swarmsConfigurable radius, speed, mass, steering weights, and priorityHeight projection can be handled through terrain, raycast, fixed altitude, or heightmap samplingIntegrationSwarm Core can run standaloneFlow Fields integrates directly with Swarm CoreSteering + Avoidance can run standalone or through Swarm Core adaptersFlow Field cost data can be used by avoidance modulesDesigned for modular extensionCompatible with custom AI decision systemsCan be combined with separate Perception and Decision modulesLimitationsFlow Fields are grid-based and primarily XZ-oriented for global navigationFlying support is suitable for layered/fixed-altitude and steering-driven vertical movement, not full volumetric 3D pathfindingNo built-in animation controllerNo built-in combat logicNo required NavMesh crowd simulationVery large scenes require tuning grid size, update rates, neighbor radius, and avoidance settings.AI tools were used to assist with code structure, documentation drafting, and implementation review. Final design decisions, testing, tuning, and asset integration were manually reviewed and refined.
