Efficient Terrain Driven Coral Coverage Using Gaussian Processes for Mosaic Synthesis

Manjanna, Sandeep; Kakodkar, Nikhil; Meghjani, Malika; Dudek, Gregory

doi:10.1109/crv.2016.63

Cited by 21 publications

(10 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Note that the expected value in (11) is nonnegative: it is a weighted sum of squared terms, where all weights are non-negative (again, due to their interpretation as probabilities). Thus, (11) is non-positive, so (10) must ≤ 0, confirming (9).…”

Section: Appendixmentioning

confidence: 84%

“…We present a proof showing that coverage of the search region to reduce probability mass is equivalent to searching for the target. We have previously demonstrated an algorithm to selectively cover a region based on an underlying reward distribution [9], [10]. This technique, however, does not scale up smoothly for larger regions because of the computational complexity.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Reinforcement Learning with Non-uniform State Representations for Adaptive Search

Manjanna

Hoof

Dudek

2018

2018 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR)

Self Cite

View full text Add to dashboard Cite

Efficient spatial exploration is a key aspect of search and rescue. In this paper, we present a search algorithm that generates efficient trajectories that optimize the rate at which probability mass is covered by a searcher. This should allow an autonomous vehicle find one or more lost targets as rapidly as possible. We do this by performing non-uniform sampling of the search region. The path generated minimizes the expected time to locate the missing target by visiting high probability regions using non-myopic path generation based on reinforcement learning. We model the target probability distribution using a classic mixture of Gaussians model with means and mixture coefficients tuned according to the location and time of sightings of the lost target. Key features of our search algorithm are the ability to employ a very general nondeterministic action model and the ability to generate action plans for any new probability distribution using the parameters learned on other similar looking distributions. One of the key contributions of this paper is the use of non-uniform state aggregation for policy search in the context of robotics.We compare the paths generated by our algorithm with other accepted spatial coverage techniques such as distribution independent boustrophedonic coverage and model dependent spiral search. We present a proof showing that rewarding for clearing probability mass instead of locating the target does not bias the objective function. The experiments show that the learned policy outperforms several well-known baselines even in scenarios different from the one it has been trained on.

show abstract

Section: Appendixmentioning

confidence: 84%

Section: Introductionmentioning

confidence: 99%

Reinforcement Learning with Non-uniform State Representations for Adaptive Search

Manjanna

Hoof

Dudek

2018

2018 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR)

Self Cite

View full text Add to dashboard Cite

show abstract

“…images of coral species [199] or to study exposure effects on coral organisms, e.g., ultraviolet radiation [200]. Generally, the best images come from coral heads that are located in more shallow waters and the best target for imaging is the most exposed coral colony.…”

Section: Discussionmentioning

confidence: 99%

Adaptive sampling of transient environmental phenomena with autonomous mobile platforms

Preston¹

2019

View full text Add to dashboard Cite

Thanks first goes to my family and friends for their support and willingness to entertain hours of conversation about robot boats and marine methane. I especially wish to acknowledge my parents for the freedom they provided me throughout my life to explore my interests, ultimately enabling me to become the independent (and quirky) person I am today. I also sincerely thank my partner, Bill, for his enthusiasm about my work, ability to engage in thoughtful conversations about robotics, and his enduring ability to make me laugh when the going gets tough. To my advisers, Anna Michel and Nicholas Roy, I extend sincere gratitude for their mentorship and insight throughout my work. I recognize the fortune I have to work with two incredible professionals in two unique fields, and the opportunities for field work and learning this has afforded. Additionally, working with their labs, WHOI DSL Michel Lab and MIT CSAIL RRG, has been a pleasure. Being surrounded by thoughtful and creative minds everyday is a privilege; and I'm especially thankful to count these individuals among my friends. I particularly thank my research collaborators Dr.Cara Manning, Dr.David Nicholson, Dr.Yogesh Girdhar, Genevieve Flaspohler, and Kevin Manganini for their insight, patience, support, and positive attitudes throughout our work together. They have taught me so much about how to be a good scientist, engineer, and communicator. As I hit this milestone on my way towards earning a PhD, I look forward to what the future has in store. I would like to dedicate this thesis to my parents, Matthew and Elizabeth Preston, to whom I owe my intellectual curiosity, confidence, and perseverance.

show abstract

“…with the objective of localizing the most exposed coral head. Coral head exposure is used to select vantage points for coral imaging [30] and in ultraviolet radiation studies on coral organisms [31]. Due to time and resource constraints, only one trial of two planners was feasible on the physical reef; we elected to demonstrate PLUMES and Boustro., one of the most canonical surveying strategies in marine sciences.…”

Section: A Bounded Convex Environmentsmentioning

confidence: 99%

Information-Guided Robotic Maximum Seek-and-Sample in Partially Observable Continuous Environments

Flaspohler

Preston

Michel

et al. 2019

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

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 continuousobservation 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.

show abstract

Efficient Terrain Driven Coral Coverage Using Gaussian Processes for Mosaic Synthesis

Cited by 21 publications

References 13 publications

Reinforcement Learning with Non-uniform State Representations for Adaptive Search

Reinforcement Learning with Non-uniform State Representations for Adaptive Search

Adaptive sampling of transient environmental phenomena with autonomous mobile platforms

Information-Guided Robotic Maximum Seek-and-Sample in Partially Observable Continuous Environments

Contact Info

Product

Resources

About