40 lines (34 loc) · 1.42 KB

Roadmap: AgentSimMiddleware (Hybrid Architecture)

Phase 0: Research & Planning

Benchmark current C# GOAP + FSM implementation
Define API for C++ core simulation
Design SoA memory layout for agents, actions, and perception
Draft adapter API for Unity integration

Phase 1: Core Simulation (C++)

Implement Agent container and update loop
Implement GOAP planner (A* / backward-chaining)
Implement FSM execution engine
LOD manager for performance scaling
WorldState representation and perception queries
Multi-threading and memory pooling

Phase 2: C# Adapter & Engine Integration

Unity adapter for world snapshots → core simulation
AgentCommand retrieval and application to GameObjects
Debug visualization: LOD, current goals, FSM states
Sample scenes with hundreds of agents
Performance profiling tools

Phase 3: Optimization & Scaling

Fine-tune multi-threaded batch sizes
Cache-friendly data access patterns
Precompute planner heuristics and action costs
Integrate optional deterministic mode for multiplayer
Performance benchmarks: 500–2000 agents at 60fps

Phase 4: Cross-Engine Expansion

Unreal Engine adapter
Godot / Stride adapter (optional)
API unification for multiple engine adapters

Phase 5: Advanced Features

Hierarchical GOAP and behavior trees
GPU-accelerated perception queries
Advanced LOD tiers (cluster-level simulation)
Networked deterministic simulation