Title:Analysis of Exploration vs. Exploitation in Adaptive Information Sampling

Abstract:Adaptive information sampling approaches enable efficient selection of mobile robot's waypoints through which accurate sensing and mapping of a physical process, such as the radiation or field intensity, can be obtained. This paper analyzes the role of exploration and exploitation in such information-theoretic spatial sampling of the environmental processes. We use Gaussian processes to predict and estimate predictions with confidence bounds, thereby determining each point's informativeness in terms of exploration and exploitation. Specifically, we use a Gaussian process regression model to sample the Wi-Fi signal strength of the environment. For different variants of the informative function, we extensively analyze and evaluate the effectiveness and efficiency of information mapping through two different initial trajectories in both single robot and multi-robot settings. The results provide meaningful insights in choosing appropriate information function based on sampling objectives.