Adaptive robot sampling improves ocean forecasts with learned models
Closed-Loop Adaptive Sampling on a Generative Ocean Model
Robotics
Summary
Collecting ocean data is hard because robots can only take so many measurements. The authors created a method that uses special computer models to predict ocean conditions and then guides robots to the most useful spots to collect data. Their approach learns from previous measurements and adapts where robots go next, making ocean monitoring more efficient. Tests showed this method usually finds better data than fixed or random sampling plans.
What this means in practice
- •For marine robotics teams: Plan efficient measurement routes for mobile ocean robots using learned forecasts to improve data collection quality and coverage.
- •For environmental monitoring agencies: Use adaptive sampling frameworks to improve sparse ocean observation programs and better track changing marine conditions.
Authors
Shayesteh Hafezi, Alan Papalia
Abstract
Adaptive sampling can substantially improve the value of sparse ocean observations by directing mobile robots toward measurements that are most informative for an evolving environment. Recent learned generative ocean models can produce high-resolution forecasts and ensembles at substantially lower computational cost than traditional numerical models, creating an opportunity to use rich predictive forecasts directly in closed-loop robotic sampling. However, these models do not natively provide the explicit uncertainty representations assumed by classical informative-planning methods, and their forecasts often evolve in learned latent spaces rather than directly in physical ocean variables. We develop a multi-robot closed-loop adaptive sampling framework that uses a single learned generative model for forecasting, uncertainty estimation, planning, and measurement assimilation. The model forecasts an ensemble in latent space, decodes it to physical ocean states, and uses ensemble statistics to construct A- or D-optimal sampling objectives. We select sampling locations subject to robot reachability under predicted currents. The collected measurements are assimilated back into the latent state using observation-conditioned diffusion sampling. Across 120 test scenarios A- and D-optimal sampling generally outperforms static, random, and systematic-coverage baselines while adding little planning overhead. Our experiments uncover several scenarios in which model uncertainty diverges from forecast error, motivating future work in uncertainty quantification.