Informative planning and online learning with sparse Gaussian processes

A big challenge in environmental monitoring is the spatiotemporal variation of the phenomena to be observed. To enable persistent sensing and estimation in such a setting, it is beneficial to have a time-varying underlying environmental model. Here we present a planning and learning method that enables an autonomous marine vehicle to perform persistent ocean monitoring tasks by learning and refining an environmental model. To alleviate the computational bottleneck caused by large-scale data accumulated, we propose a framework that iterates between a planning component aimed at collecting the most information-rich data, and a sparse Gaussian Process learning component where the environmental model and hyperparameters are learned online by taking advantage of only a subset of data that provides the greatest contribution. Our simulations with ground-truth ocean data shows that the proposed method is both accurate and efficient.

Authors

• Kai-Chieh Ma
• Lantao Liu
• Gaurav S. Sukhatme

Keywords

• Data models
• Planning
• Training
• Computational modeling
• Robot sensing systems
• Predictive models
• Kernel
• Online Learning
• Online Information
• Gaussian Process
• Sparse Gaussian Process
• Subset Of Data
• Planning Methods
• Ocean Data
• Component Of Planning
• Sparse Component
• Training Set
• Covariance Matrix
• Mutual Information
• Kernel Function
• Kullback-Leibler
• Information Gain
• Leave-one-out Cross-validation
• Dynamic Programming
• Kriging
• Path Planning
• Prediction Map
• Autonomous Underwater Vehicles
• Salinity Data
• Sample Spot
• Grid Map
• Traveling Salesman Problem
• Subset Of Points