Informed Sampling and Adaptive Monitoring using Sparse Gaussian Processes

Physical sample collection from the hotspots of an environmental field is of interest to scientists, environmental protection agencies as well as public utility authorities. Such a sample collection task requires prior knowledge of the locations of the hotspots, which is generally not available. In this paper, we suggest an algorithm, Sampling and Adaptive Monitoring (SAM), to perform both the tasks of approximating the environmental field and sampling from the hotspots, simultaneously. We test our performance using a user defined utility function and show that we are able to obtain a good approximation of the field and collect samples from the hotspots within the stipulated time.