Sea Lock: Predictive Multi-Agent Wargaming for Adversarial Naval Blockade and Command-and-Control

Mission-level Digital Twins increasingly support defence decision-making, yet most existing deployments model friendly forces and environments with high fidelity while treating adversarial behaviour as scripted or scenario-fixed — limiting their value as anticipatory Command-and-Control (C2) tools in contested maritime settings where red and blue forces co-evolve under uncertainty and partial observability.

We present Sea Lock, a multi-agent reinforcement learning framework that introduces an adaptive adversarial learning layer into mission-space Digital Twins. The blockade engagement is formalised as a partially observable stochastic game over heterogeneous surface vessels, jointly modelling formation, reconnaissance, electronic warfare, and salvo combat.

Three training regimes are compared under a centralised-training, decentralised-execution architecture: standard PPO, fictitious self-play, and an LLM-shaped doctrinal reward variant. A lightweight Recurrent State-Space Model, trained on interaction trajectories, enables counterfactual rollouts of alternative manoeuvre doctrines without re-simulating the environment.

Sea Lock is the maritime rendering of a broader adversarial- simulation engine designed for cross-domain composition. The same engine architecture extends to land logistics, air swarm operations, cyber resilience, and integrated multi-domain C2 — each rendering shares the engine while varying observation and action spaces, reward composition, and domain-specific physics. Digital Twin is one application; predictive wargaming, force structure analysis, doctrine red-teaming, and operator decision support are others.