Agents & Scenarios Architecture

This section documents the multi-agent taxonomy, built-in urban scenario benchmarks, and BenchMARL integration APIs.

• Agent Taxonomy & State Spaces
  • Agent Taxonomy
  • Multi-Agent Group Mapping (group_map)
• Urban Scenario Benchmarks
  • Scenario Overview
  • 1. UAV Navigation (uav_navigation)
  • 2. UAV-UE LoS Link Maintenance (uav_ue_los)
  • 3. 3D Spatial Area Coverage (coverage)
  • 4. UAV-MEC Task Offloading (uavmec_offloading)
  • 5. Dynamic Ground User Mobility (uav_mobile_ue)
  • 6. Autonomous LiDAR Navigation (uav_lidar_navigation)
  • 7. High-Fidelity Physics & Propagation (uavmec_advanced_physics)
  • 8. Default Scenario (default)
  • Dynamic Scenario Auto-Registration
• BenchMARL Integration Guide
  • Architecture Overview
  • Code Example: Training MADDPG on Urban Navigation
  • Code Example: MAPPO on MEC Offloading
  • Code Example: MADDPG on High-Fidelity Physics
  • Code Example: MASAC on LiDAR Urban Canyon Navigation
  • Evaluating & Plotting Results