Surface Metering & IG-RL

Graph-based reinforcement learning for multi-agent aircraft surface movement optimisation.

Intelligent Graph Reinforcement Learning (IG-RL) models departure aircraft as decentralised agents on an airport surface graph. Each agent learns to optimise its own taxi trajectory while collaborating to reduce airport-wide congestion, subject to operational constraints such as separation minima and route precedence.

What this demonstrates:

  • Multi-agent reinforcement learning
  • Graph search and path planning
  • Discrete-event simulation and scenario analysis
  • Network optimisation for transport systems
  • Integration with real operational data