Autonomous oil spill response through liquid neural trajectory modeling and coordinated marine robotics

Marine oil spills pose grave environmental and economic risks, threatening marine ecosystems, coastlines, and dependent industries. Predicting and managing oil spill trajectories is highly complex, due to the interplay of physical, chemical, and environmental factors such as wind, currents, and temperature, which makes timely and effective response challenging. Accurate real-time trajectory forecasting and coordinated mitigation are vital for minimizing the impact of these disasters. This study introduces an integrated framework combining Liquid Time-Constant Networks (LTCNs) with multi-agent swarm robotics for real-time oil spill trajectory prediction and coordinated response. Our approach implements three complementary LTC solver variants optimized for different operational scenarios: RK4 for critical emergency response, Explicit for operational monitoring, and Euler for large-scale surveillance. The framework is validated using Deepwater Horizon satellite observations under moderate sea state conditions where Loop Current advection and wind forcing dominated transport. Results demonstrate superior spatial prediction accuracy (IoU 0.82-0.84), significantly surpassing Transformer (0.71) and LSTM (0.68) baselines. Crucially, the LTC model maintains realistic irregular boundary geometries (64+ vertices, complexity ratios 0.89-0.96) compared to oversimplified baseline predictions (5-12 vertices, complexity 0.48-0.61) that exhibit unrealistic circular approximations. The framework's integration with MOOS-IvP enables autonomous fleet coordination, demonstrating scalable, fault-tolerant response capabilities. This work advances physics-based environmental prediction while providing operational flexibility through solver-specific deployment strategies.