Data‐driven learning and planning for environmental sampling

Abstract

Robots such as autonomous underwater vehicles (AUVs) and autonomous surface vehicles (ASVs) have been used for sensing and monitoring aquatic environments such as oceans and lakes. Environmental sampling is a challenging task because the environmental attributes to be observed can vary both spatially and temporally, and the target environment is usually a large and continuous domain whereas the sampling data is typically sparse and limited.The challenges require that the sampling method must be informative and efficient enough to catch up with the environmental dynamics. In this paper we present a planning and learning method that enables a sampling robot to perform persistent monitoring tasks by learning and refining a dynamic "data map" that models a spatiotemporal environment attribute such as ocean salinity content. Our environmental sampling framework consists of two components: to maximize the information collected, we propose an informative planning component that efficiently generates sampling waypoints that contain the maximal information; To alleviate the computational bottleneck caused by large-scale data accumulated, we develop a component based on a sparse Gaussian Process whose hyperparameters are learned online by taking advantage of only a subset of data that provides the greatest contribution. We validate our method with both simulations running on real ocean data and field trials with an ASV in a lake environment. Our experiments show that the proposed framework is both accurate in learning the environmental data map and efficient in catching up with the dynamic environmental changes † .

Off‐road ground robot path energy cost prediction through probabilistic spatial mapping

“…Thus, satellite imagery or other spatial maps of the area would be particularly useful in helping to increase the length-scale through detection of known transitions between terrains. Furthermore, numerous effective strategies exist for managing the computational cost of GPR, including using local approximations (Vasudevan, Ramos, Nettleton, & Durrant-Whyte, 2009), achieving sparsity through subsampling the data set (Ma, Liu, Heidarsson, & Sukhatme, 2018) or Hilbert space approximations (Solin, Kok, Wahlström, Schön, & Särkkä, 2018).…”

Adaptive sampling of transient environmental phenomena with autonomous mobile platforms

“…For complex phenomenon, GPs may not be expressive enough or simply may not have features that can be described using a Gaussian relationship. To improve the flexibility of GPs, kernel learning [203,204] is one potential avenue for development, in which using observations collected online or through a training dataset offline, the hyperparameters of a kernel function can be optimized to better conform the kernel to the relationships inherent in the data. However, alternative representations for some environmental phenomenon will still be required.…”

Sensing water properties at precise depths from the air

“…Several efforts (Alam, Reis, Bobadilla, & Smith, ; Das et al., ; Hollinger and Sukhatme, ; Lawrance, Chung, & Hollinger, ; Ma, Liu, Heidarsson, & Sukhatme, ; Yoo et al., ) use an information quality metric (e.g., information gain or variance reduction) along with system constraints (e.g., battery) to plan motion during environmental monitoring in aqueous environments, mostly with AUVs and ASVs. These works plan a path and choose a waypoint (or depth) based on a data‐driven metric, whereas our work assumes the target depth has already been chosen.…”