Authors: Genevieve E Flaspohler (MIT); Victoria Preston (MIT); Anna Michel (Woods Hole Oceanographic Institution); Yogesh Girdhar (WHOI); Nicholas Roy (MIT)
Abstract: We present PLUMES, a planner for localizing and collecting samples at the global maximum of an a priori unknown and partially observable continuous environment. This "maximum seek-and-sample" (MSS) problem is pervasive in the environmental and earth sciences. Experts want to collect scientifically valuable samples at an environmental maximum (e.g., an oil-spill source), but do not have prior knowledge about the phenomenon's distribution. We formulate the MSS problem as a partially-observable Markov decision process (POMDP) with continuous state and observation spaces, and a sparse reward signal. To solve the MSS POMDP, PLUMES uses an information-theoretic reward heuristic with continuous-observation Monte Carlo Tree Search to efficiently localize and sample from the global maximum. In simulation and field experiments, PLUMES collects more scientifically valuable samples than state-of-the-art planners in a diverse set of environments, with various platforms, sensors, and challenging real-world conditions.
Authors: Youngsun Kwon (KAIST); Bochang Moon (Gwangju Institute of Science and Technology); Sungeui Yoon (KAIST)
Abstract: In this paper, we present a new approach, AKIMap, that uses an adaptive kernel inference model for dense and sharp occupancy representations. Our approach is based on the multivariate kernel estimation, and we propose a simple, two-step based approach that selects an adaptive bandwidth matrix for an efficient and accurate occupancy estimation. To utilize correlations of occupancy observations such given sparse and non-uniform distributions of point samples, we propose to use the covariance matrix as an initial bandwidth matrix, and then optimize the bandwidth matrix by adjusting its scale in an efficient way for on-the-fly mapping. We test the practical benefit and robustness in occupancy estimation of the proposed method using a synthetic dataset. In various experiments including equal-data or equal-time settings, our adaptive inference model robustly shows a higher estimation accuracy over the state-of-the-art method.
Authors: Nare Karapetyan (University of South Carolina); Ioannis Rekleitis (University of South Carolina)
Abstract: Autonomous area coverage is an important problem arising in environmental monitoring, search and rescue,infrastructure monitoring, and survey operations. Particularly when an area of interest has unstructured boundaries, such as rivers, the state-of the art approaches of area coverage path planning are not directly applicable. In addition, during coverage operations, and path planning in general, in dynamic environments inaccurate navigation from waypoint to waypoint presents extra challenges. In this work we present an overview of different patterns that can be utilized for river coverage in surveying, water sampling, and bathymetric mapping operations. In addition we discuss how a reactive method can be used for more accurate navigation when the environmental forces affect the navigation of the surface vehicle. Described methods have been tested utilizing an autonomous surface vehicle on different segments of the Congaree River in South Carolina, USA. All experiments resulted in total of more than 55km of coverage trajectories in the field.
Authors: Martin J Schuster (German Aerospace Center (DLR))*; Marcus Müller (German Aerospace Center (DLR)); Sebastian Brunner (German Aerospace Center (DLR)); Hannah Lehner (German Aerospace Center (DLR)); Peter Lehner (German Aerospace Center (DLR)); Andreas Dömel (German Aerospace Center (DLR)); Mallikarjuna Vayugundla (German Aerospace Center (DLR)); Florian Steidle (German Aerospace Center (DLR)); Philipp Lutz (German Aerospace Center (DLR)); Ryo Sakagami (German Aerospace Center (DLR)); Lukas Meyer (German Aerospace Center (DLR)); Rico Belder (German Aerospace Center (DLR)); Michal Smisek (German Aerospace Center (DLR)); Wolfgang Stürzl (German Aerospace Center (DLR)); Rudolph Triebel (German Aerospace Center (DLR)); Armin Wedler (German Aerospace Center (DLR))
Abstract: Teams of mobile robots will play a crucial role in future scientific missions to explore the surfaces of extraterrestrial bodies such as Moon or Mars. Taking scientific samples is an expensive task when operating far away in challenging, previously unknown environments, especially in hard-to-reach areas, such as craters, pits, and subterranean caves. In contrast to current single-robot missions, future robotic teams will increase efficiency via increased autonomy and parallelization, improve robustness via functional redundancy, as well as benefit from complementary capabilities of the individual robots. In this work, we present our heterogeneous robotic team consisting of flying and driving robots that we plan to deploy on a scientific sampling demonstration mission in a Moon-analogue environment on Mt. Etna, Sicily, Italy in 2020 as part of the ARCHES project. We first describe the robots' individual capabilities and then highlight their tasks in the joint mission scenario. In addition, we present preliminary experiments on important subtasks: the analysis of volcanic rocks via spectral images, collaborative multi-robot 6D SLAM in a Moon-analogue environment as well as with a rover and a drone in a Mars-like scenario, and demonstrations of autonomous robotic sample-return missions therein.
Authors: Alberto Quattrini Li (Dartmouth College); Holly Ewing (Bates College); Annie Bourbonnais (University of South Carolina); Paolo Stegagno (University of Rhode Island); Ioannis Rekleitis (University of South Carolina); Denise Bruesewitz (Colby College); Kathryn Cottingham (Dartmouth College); Devin Balkcom (Dartmouth); Mark Ducey (University of New Hampshire); Kenneth Johnson (University of New Hampshire); Stephen Licht (University of Rhode Island); David Lutz (Dartmouth College); Jason O'Kane (University of South Carolina); Michael Palace (University of New Hampshire); Christopher Roman (University of Rhode Island); V. S. Subrahmanian (Dartmouth College, Hanover); Kathleen Weathers (Cary Institute of Ecosystem Studies)
Abstract: This extended abstract describes a joint effort to model and predict harmful cyanobacterial blooms in lakes of an interdisciplinary team with expertise in big data, environmental science, ecology, human demography, instrumentation, and robotics from four states: Maine, New Hampshire, Rhode Island, and South Carolina. This project uniquely integrates current methodology for data collection, including remote sensing and manual limnological sampling, together with heterogeneous robotic and sensor systems to extend the spatial and temporal sampling. Such big amount of data will be analyzed and processed using ensemble prediction models for determining the development and severity of blooms both in time and space (when and where) and for testing limnological hypotheses. While this project just started and does not have new result yet, this paper provides insights on open research questions and the methodology used, as well as best practices for interdisciplinary collaboration across different departments, institutions, and citizen scientists.
Authors: Robert F Codd-Downey (York University); Michael Jenkin (York University); Bir Bikram Dey (York University)
Abstract: Invasive aquatic plant species pose a major threat to domestic flora and fauna and can in turn negatively impact local economies. Numerous strategies have been developed to harvest and remove these plant species from the environment. However it is still an open question as to which method is best suited to removing a particular invasive species. Here we detail efforts to monitor plant growth and evaluate harvesting methods using an autonomous surface vehicle (ASV) outfitted with environmental and positioning sensors.
Authors: Larkin L Heintzman (Virginia Tech); Barnabas Gavin Cangan (Virginia Tech); Amanda Hashimoto (Virginia Tech); Ryan Williams (Virginia Tech); Nicole Abaid (Virginia Tech)
Abstract: In this work, our goal is to extend the existing search and rescue paradigm by allowing teams of autonomous unmanned aerial vehicles (UAVs) to collaborate effectively with human searchers on the ground. We derive a framework that includes a simulated lost person behavior model, as well as a human searcher behavior model that is informed by data collected from past search tasks. These models are used together to create a probabilistic heatmap of the lost person's position and anticipated searcher trajectories. We then use Gaussian processes with a Gibbs' kernel to accurately model a limited field-of-view (FOV) sensor, e.g., thermal cameras, from which we derive a risk metric that drives UAV path optimization. Our framework finally computes a set of search paths for a team of UAVs to autonomously complement human searchers' efforts.
Back ↩