Source author record

Victoria Preston

Victoria Preston appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Robotics

Catalog footprint

What is connected

3works

1topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2022arXiv

Enabling Under-Ice Geochemical Observations with a Size, Weight, and Power-Constrained Robot

Estimates of greenhouse gas emissions from Arctic estuarine environments are dominated by in situ summer-time ice-free dissolved gas measurements due to the logistical ease of performing field observations in these conditions. Recent evidence in coastal Arctic environments, however, has demonstrated that dissolved methane (CH4) and carbon dioxide (CO2) are strongly seasonally variable, and at least one significant gas ventilation event occurs during the spring freshet. Whether the Arctic serves as a source or sink of greenhouse emissions has significant implications on modeling climate change and its feedback mechanisms. To enable higher resolution spatiotemporal measurements of dissolved gases in typically undersampled conditions, remotely operated vehicles (ROVs) can be used to extract near continuous water samples below ice before and during the spring freshet. Here, we present a size, weight, and power-constrained (SWAP) underwater vehicle (UV) and novel geochemical sampling system suitable for taking under-ice geochemical observations and demonstrate the proposed system in a field-analog setting for Arctic estuarine studies.

preprint2022arXiv

Robotic Planning under Uncertainty in Spatiotemporal Environments in Expeditionary Science

In the expeditionary sciences, spatiotemporally varying environments -- hydrothermal plumes, algal blooms, lava flows, or animal migrations -- are ubiquitous. Mobile robots are uniquely well-suited to study these dynamic, mesoscale natural environments. We formalize expeditionary science as a sequential decision-making problem, modeled using the language of partially-observable Markov decision processes (POMDPs). Solving the expeditionary science POMDP under real-world constraints requires efficient probabilistic modeling and decision-making in problems with complex dynamics and observational models. Previous work in informative path planning, adaptive sampling, and experimental design have shown compelling results, largely in static environments, using data-driven models and information-based rewards. However, these methodologies do not trivially extend to expeditionary science in spatiotemporal environments: they generally do not make use of scientific knowledge such as equations of state dynamics, they focus on information gathering as opposed to scientific task execution, and they make use of decision-making approaches that scale poorly to large, continuous problems with long planning horizons and real-time operational constraints. In this work, we discuss these and other challenges related to probabilistic modeling and decision-making in expeditionary science, and present some of our preliminary work that addresses these gaps. We ground our results in a real expeditionary science deployment of an autonomous underwater vehicle (AUV) in the deep ocean for hydrothermal vent discovery and characterization. Our concluding thoughts highlight remaining work to be done, and the challenges that merit consideration by the reinforcement learning and decision-making community.

preprint2020arXiv

Enabling Topological Planning with Monocular Vision

Topological strategies for navigation meaningfully reduce the space of possible actions available to a robot, allowing use of heuristic priors or learning to enable computationally efficient, intelligent planning. The challenges in estimating structure with monocular SLAM in low texture or highly cluttered environments have precluded its use for topological planning in the past. We propose a robust sparse map representation that can be built with monocular vision and overcomes these shortcomings. Using a learned sensor, we estimate high-level structure of an environment from streaming images by detecting sparse vertices (e.g., boundaries of walls) and reasoning about the structure between them. We also estimate the known free space in our map, a necessary feature for planning through previously unknown environments. We show that our mapping technique can be used on real data and is sufficient for planning and exploration in simulated multi-agent search and learned subgoal planning applications.