Information-Based Hierarchical Planning for a Mobile Sensing Network in Environmental Mapping

Abstract

This article investigates the problem of informationbased sampling design and path planning for a mobile sensing network to predict scalar fields of monitored environments. A hierarchical framework with a built-in Gaussian Markov random field model is proposed to provide adaptive sampling for efficient field reconstruction. In the proposed framework, a nonmyopic planner is operated at a sink to navigate the mobile sensing agents in the field to the sites that are most informative. Meanwhile, a myopic planner is carried out on board each agent. A tradeoff between computationally intensive global optimization and efficient local greedy search is incorporated into the system. The mobile sensing agents can be scheduled online through an anytime algorithm to visit and observe the high-information sites. Experiments on both synthetic and real-world datasets are used to demonstrate the feasibility and efficiency of the proposed planner in model exploitation and adaptive sampling for environmental field mapping. Index Terms-Adaptive sampling, environmental field mapping, Gaussian Markov random fields (GMRFs), information-driven planning, mobile sensing networks (MSNs). I. INTRODUCTIONM OBILE sensing networks (MSNs) can provide unprecedented flexibility, efficiency, and effectiveness in information gathering. They can improve the performance of an environmental monitoring process. With the support of mobile sensing, measurements of the monitored variables over both spatial and temporal scales can be collected to estimate, interpret, and reconstruct the environmental field of interest. Due to these advantages, MSNs have been developed and deployed to provide in situ measurements and field maps in many environmental monitoring programs [1]-[3].When implementing an MSN, owing to the limited number and mobility of the sensing agents (e.g., mobile sensors or T. Li and C. W. de Silva are with the

Multistep Predictions for Adaptive Sampling in Mobile Robotic Sensor Networks Using Proximal ADMM

“…Furthermore, the multi-robot coverage control has been taken into account simultaneously with the sampling design in [27], [28]. Other metrics having been utilized in adaptive sampling are based on information criteria, including Fisher information matrix, entropy, and mutual information [29]- [31]. In [32], the upper confidence bound of a GP was adopted in an optimization objective and the cross-entropy method was used to optimize the sampling paths.…”

Multi-Step Predictions for Adaptive Sampling using Proximal ADMM

“…Another metric having been utilized in deriving the adaptive sampling optimization problem is relied on the information criteria including Fisher information matrix, entropy and mutual information [30]- [32]. In [33], the upper confidence bound of the GP is adopted in an optimization criterion and the crossentropy method is used to optimize the sampling paths.…”

Multi-Step Predictions for Adaptive Sampling using Proximal ADMM

“…Furthermore, the multi-robot coverage control has been taken into account simultaneously with the sampling design in [28], [29]. Other metrics having been utilized in adaptive sampling are based on information criteria, including Fisher information matrix, entropy, and mutual information [30]- [32]. In [33], the upper confidence bound of a GP was adopted in an optimization objective and the cross-entropy method was used to optimize the sampling paths.…”