A Bayesian Optimization Approach for Multi-Function Estimation for Environmental Monitoring Using an Autonomous Surface Vehicle: Ypacarai Lake Case Study

Bayesian optimization is a sequential method that can optimize single and costly objective function based on surrogate model. In this work, we propose system dedicated to monitoring estimating multiple water quality parameters simultaneously using autonomous surface vehicle. The proposed work combines different strategies methods for task, evaluating two approaches acquisition fusion: the coupled decoupled techniques. We also consider dynamic parametrization of maximum measurement distance traveled by ASV so balances total number measurements distance, which related energy required. To evaluate approach, Ypacarai Lake (Paraguay) serves as test scenario, where maps parameters, such pH dissolved oxygen, need be obtained efficiently. compared with predictive entropy search multi-objective constraints (PESMOC) algorithm genetic (GA) path planning scenario. results show approach 10.82% better than other in terms R2 score noiseless up 17.23% when data are noisy. Additionally, achieves good average computational time whole mission methods, 3% GA technique 46.5% PESMOC approach.