Content uploaded by Kanna Rajan
Author content
All content in this area was uploaded by Kanna Rajan on Mar 11, 2015
Content may be subject to copyright.
A preview of the PDF is not available
Exploring Space-Time Tradeoffs in Autonomous Sampling for Marine Robotics,
Abstract and Figures
In the coastal ocean, biological and physical dynamics vary on spatiotemporal scales spanning many orders of magnitude. At large spatial (O(100km)) and temporal (O(weeks to months)) scales, traditional shipboard and moored measurements are very effective at quantifying mean and varying oceanic properties. At scales smaller than the internal Rossby radius (O(10km) for typical coastal stratification at mid-latitude), horizontal, vertical and temporal inhomogeneity is the rule rather than the exception.
Figures - uploaded by Kanna Rajan
Author content
All figure content in this area was uploaded by Kanna Rajan
Content may be subject to copyright.
... This was developed further in (21), where the sampling resolution was modified by adjusting the undulation angle of a glider while accounting for ocean currents, with the aim to characterize and sample blooms. Graham et al. (22) discussed problems related to estimating spatiotemporal correlation in time-varying fields and compared Euclidean and Lagrangian approaches using AUV and drifter information to perform spatial interpolation. Feature tracking of patches and plumes was also discussed in (23), where a plan-based policy was learned, tested, and evaluated using simulated patches of Chla and then sampled in 2D in the open ocean. ...
... We used a spatial statistical model in the form of a GP to model the SCM. The GP formulation is widely used for dealing with spatial applications in the ocean (22,53). Its popularity is largely due to the simple yet flexible modeling formulation, which allows consistent and efficient data conditioning. ...
... Ideally, the AUV should therefore try to "stay" with the same water mass, working in a Lagrangian frame of reference to reduce these effects. Lagrangian correction is, however, nontrivial, and, unless a good proxy measurement of advection can be provided (such as from a surface drifter), there is limited value in adding to experimental complexity [see (22) for a more detailed discussion]. A simple and effective measure to account for this is to limit the method/survey area to subkilometer size, setting a bound on these uncertainties similar to the enclosure criterion used in (9). ...
Currents, wind, bathymetry, and freshwater runoff are some of the factors that make coastal waters heterogeneous, patchy, and scientifically interesting—where it is challenging to resolve the spatiotemporal variation within the water column. We present methods and results from field experiments using an autonomous underwater vehicle (AUV) with embedded algorithms that focus sampling on features in three dimensions. This was achieved by combining Gaussian process (GP) modeling with onboard robotic autonomy, allowing volumetric measurements to be made at fine scales. Special focus was given to the patchiness of phytoplankton biomass, measured as chlorophyll a (Chla), an important factor for understanding biogeochemical processes, such as primary productivity, in the coastal ocean. During multiple field tests in Runde, Norway, the method was successfully used to identify, map, and track the subsurface chlorophyll a maxima (SCM). Results show that the algorithm was able to estimate the SCM volumetrically, enabling the AUV to track the maximum concentration depth within the volume. These data were subsequently verified and supplemented with remote sensing, time series from a buoy and ship-based measurements from a fast repetition rate fluorometer (FRRf), particle imaging systems, as well as discrete water samples, covering both the large and small scales of the microbial community shaped by coastal dynamics. By bringing together diverse methods from statistics, autonomous control, imaging, and oceanography, the work offers an interdisciplinary perspective in robotic observation of our changing oceans.
... Generally, knowledge is not readily available, incomplete, and include signicant uncertainty. Driven by these factors, the demand for adaptive and more advanced autonomy concepts is contemplated across a wide range of autonomous research [9,33,19,16,46,26]. The notion of dealing with; autonomy, adaptation/re-planning, and computational search between dierent action outcomes, are ideas central to the elds of articial intelligence (AI) and machine learning (ML). ...
... They can be deployed on the surface or under water. One example of use is given in [19], where cooperation with AUVs have been demonstrated for helping the AUV "stay" ...
... Incorporating adaptive algorithms into mobile agents have been researched extensively and several approached have been investigated, see e.g. [9,33,19,16,46]. The ability to adjust execution on the basis of sensory information is a problem not just in robotics, but also statistics, geology, atmospheric science, and other elds concerned with optimizing information gathering. ...
The backdrop for this review is rooted in ocean observation and monitoring, using machine intelligence to extend autonomous underwater vehicle (AUV) data collection capabilities. AUVs have provided scientists with a powerful tool for oceanographic research, and have changed the way ocean science is conducted, but the potential for further innovation is still great. The drive for this is twofold; sustainability and management of natural resources is regularly cited as the biggest problem of our generation. The ocean plays the central role as both a resource and early indicator for climate change, and the AUV is the tool which is best suited to provide the facts necessary for decision makers. Secondly, data collection at sea is still a challenging and an expensive enterprise; most AUV control agent systems today rely on a pre-programmed plan (Mcgann, 2007), consisting of sequential behaviors scripted with mission planning tools. This notion is not only restricted to the field of marine robotics, but information gathering platforms in general. Creative and novel solutions are therefore imperative, and constitute an excellent field for research.
... There is a large body of literature on maximizing information gain from in situ measurements to characterize phenomena, or providing estimation of a scalar field. Zhang and Sukhatme (2007) showed adaptive sampling schemes for reconstructing a temperature field using a sensor network of both static and mobile sensors, while Graham et al. (2012) discussed the use of Gaussian processes (GPs) and the problems relating to environment reconstruction in the ocean with different correlation kernels. Yilmaz, Evangelinos, Lermusiaux, and Patrikalakis (2008) used a mixed integer programming utility combining reduction of uncertainty and physical constraints. ...
... Traditional techniques, like shipboard and moored measurements, can be effective at large spatial (O(100 km)) and temporal (O(week to months)) scales, but have proved difficult for submesoscale (smaller than an internal Rossby radius of (O(10 km))) variability (Graham et al., 2012). The importance of these dynamics for physical ocean processes is significant (Barth, Hebert, Dale, & Ullman, 2004) and directly influences primary production (Lévy, 2003) and patch formation (Franks, 1992) of biological signatures. ...
... The focus of the statistical model applied in this work is to approximate the underlying distribution of ocean temperature, specified from hindcast data from SINMOD. Using a GP to model temperature as a spatial phenomenon has been studied before (e.g., Cressie & Wikle, 2011;Graham et al., 2012). Based on the characteristics of our ocean model data, the GP is a reasonable model to use for temperatures, as no heavy tails or skewness was significant in the temperature data used for modeling. ...
Efficient sampling of coastal ocean processes, especially mechanisms such as upwelling and internal waves and their influence on primary production, is critical for understanding our changing oceans. Coupling robotic sampling with ocean models provides an effective approach to adaptively sample such features. We present methods that capitalize on information from ocean models and in situ measurements, using Gaussian process modeling and objective functions, allowing sampling efforts to be concentrated to regions with high scientific interest. We demonstrate how to combine and correlate marine data from autonomous underwater vehicles, model forecasts, remote sensing satellite, buoy, and ship‐based measurements, as a means to cross‐validate and improve ocean model accuracy, in addition to resolving upper water‐column interactions. Our work is focused on the west coast of Mid‐Norway where significant influx of Atlantic Water produces a rich and complex physical–biological coupling, which is hard to measure and characterize due to the harsh environmental conditions. Results from both simulation and full‐scale sea trials are presented.
... The T-REX [47,48] was developed by the Autonomy Group at Monterey Bay Aquarium Research Institute (MBARI), and has been validated in several real case scenarios. For instance, it has been successfully used in some of the experiments of the Controlled, Agile, and Novel Ocean Network (CANON) initiative at MBARI [49,50], as well as other marine and service robotics experiments [45] and simulations [51]. It also had an executive implementation as a node part of ROS up until the C Turtle release [52]. ...
Almost every research project that focuses on the cooperation of autonomous robots for underwater operations designs their own architectures. As a result, most of these architectures are tightly coupled with the available robots/vehicles for their respective developments, and therefore the mission plan and management is done using an ad-hoc solution. Typically, this solution is tightly coupled to just one underwater autonomous vehicle (AUV), or a restricted set of them selected for the specific project. However, as the use of AUVs for underwater operations increases, there is the need to identify some commonalities and weaknesses of these architectures, specifically in relation to mission planning and management. In this paper, we review a selected number of architectures and frameworks that in one way or another make use of different approaches to mission planning and management. Most of the selected works were developed for underwater operations. Still, we have included some other architectures and frameworks from other domains that can be of interest for the survey. The explored works have been assessed using selected features related to mission planning and management, considering that underwater operations are performed in an uncertain and unreliable environment, and where unexpected events are not strange. Furthermore, we have identified and highlighted some potential challenges for the design and implementation of mission managers. This provides a reference point for the development of a mission manager component to be integrated in architectures for cooperative robotics in underwater operations, and it can serve for the same purposes in other domains of application.
... Fig. 36 shows one such experiment in September 2011 with a GPS-enabled Wirewalker was tracked by the Dorado running T-REX. We are currently investigating spatio-temporal correlations [144] of CTD data obtained by the Wirewalker and tying it to the observations made in the contextual environment by Dorado's sensors [145]. ...
Introduction to Autonomy for Marine Robots.- Autonomy for Unmanned Marine Vehicles with MOOS-IvP.- Towards Deliberative Control in Marine Robotics.- Path Planning for Autonomous Underwater Vehicles.- An Ontology Based Approach to Fault Tolerant Mission Execution for Autonomous Platforms.- Cooperation Between Underwater Vehicles.- Behaviour Adaptation by Means of Reinforcement Learning.- Simultaneous Localization and Mapping in Marine Environments.
... Generative planning is done using the Europa system (Frank and Jonsson 2003) with a timeline-based modelling language. T-REX has been used successfully to plan and execute underwater AUV missions at the Monterey Bay Aquarium Research Institute (MBARI) (Graham et al. 2012;Py, Rajan, and McGann 2010;Magazzeni et al. 2014). Task planning has also been embedded into robotic systems in a number of other ways. ...
The Robot Operating System (ROS) is a set of software libraries and tools used to build robotic systems. ROS is known for a distributed and modular design. Given a model of the environment, task planning is concerned with the assembly of actions into a structure that is predicted to achieve goals. This can be done in a way that minimises costs, such as time or energy. Task planning is vital in directing the actions of a robotic agent in domains where a causal chain could lock the agent into a dead-end state. Moreover, planning can be used in less constrained domains to provide more intelligent behaviour. This paper describes the ROSPlan framework, an architecture for embedding task planning into ROS systems. We provide a description of the architecture and a case study in autonomous robotics. Our case study involves autonomous underwater vehicles in scenarios that demonstrate the flexibility and robustness of our approach. Copyright © 2015, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.
... Zhang et al. (2012b,a) carried out atsea experiments where measurements both drove trajectory decisions and triggered collection of large samples. A single vehicle has successfully tracked a plankton bloom (Godin et al., 2011), while a coordinated approach for a similar problem using a drifter and vehicle has been studied in Das et al. (2012) and Graham et al. (2013). A collaborative control technique for tracking Lagrangian coherent structures is presented in Michini et al. (2014), and a distributed approach for plume and thermocline tracking is considered in Petillo et al. (2012). ...
We present an integrated framework for joint estimation and pursuit of dynamic features in the ocean, over large spatial scales and with multiple collaborating vehicles relying on limited communications. Our approach uses ocean model predictions to design closed-loop networked control at short time scales, and the primary innovation is to represent model uncertainty via a projection of ensemble forecasts into local linearized vehicle coordinates. Based on this projection, we identify a stochastic linear time-invariant model for estimation and control design. The methodology accurately decomposes spatial and temporal variations, exploits coupling between sites along the feature, and allows for advanced methods in communication-constrained control. Simulations with three example datasets successfully demonstrate the proof-of-concept.
The study of dynamic features of the ocean, in which complex physical, chemical, and biological interactions evolve on multiple time scales, poses significant sampling challenges because the required spatial and temporal resolutions are not possible by ship or satellite studies alone. Satellite remote sensing captures only surface effects while expensive research vessels can only make discrete observations in finite periods of time. Our work with networked marine robotics in the aerial, surface, and underwater domains is at the vanguard of a new approach to scientific exploration and observation, which brings together several technologies to enable oceanographic vessels and robots to work in tandem, thus expanding the observational footprint of these vessels. We describe a scientific cruise in the Spring of 2018 in the open waters of the Pacific where we deployed a fleet of autonomous robots to demonstrate this approach for the synoptic observation of mesoscale and sub-mesoscale features of a frontal zone. We articulate the elements and methods to multi-vehicle coordination and challenges that lie ahead in ocean observation.
Fronts between Arctic- and Atlantic-origin waters are characterized by strong lateral gradients in temperature and salinity. Ocean processes associated with fronts are complex with considerable space and time variability. Therefore, resolving the processes in frontal zones by observation is challenging but important for understanding the associated physical–biological interactions and their impact on the marine ecosystem. The use of autonomous robotic vehicles and in situ data-driven sampling can help improve and augment the traditional sampling practices, such as ships and profiling instruments. Here, we present the development and results of using an autonomous agent for detection and sampling of an Arctic front, integrated on board an autonomous underwater vehicle. The agent is based on a subsumption architecture implemented as behaviors in a finite-state machine. Once a front is detected, the front tracking behavior uses observations to continuously adapt the path of the vehicle to perform transects across the front interface. Following successful sea trials in the Trondheimsfjord, the front-tracking agent was deployed to perform a full-scale mission near 82
$^{\circ }$
N north of Svalbard, close to the sea ice edge. The agent was able to detect and track an Arctic frontal feature, performing a total of six crossings while collecting vertical profiles in the upper 90 m of the water column. Measurements yield a detailed volumetric description of the frontal feature with high resolution along the frontal zone, augmenting ship-based sampling that was run in parallel.
In this chapter Teleo-Reactive Executive (T-REX) is designed, developed, tested and deployed as an onboard adaptive control system that integrates artificial intelligence (AI)-based planning and probabilistic state estimation in a hybrid executive. Probabilistic state estimation integrates a number of science observations to produce a likelihood that the vehicle sensors perceive a feature of interest. Onboard planning and execution enable adaptation of navigation and instrument control based on the probability of having detected such a phenomenon. It further enables goal-directed commanding within the context of projected mission state and allows for replanning for off-nominal situations and opportunistic science events.
We present AUV survey methodologies to track and sample an advecting patch of water. Current AUV-based sampling rely primarily on geographic waypoint track-line surveys that are suitable for static or slowly changing features. When studying dynamic, rapidly evolving oceanographic features, such methods at best introduce error through insufficient spatial and temporal resolution, and at worst completely miss the spatial and temporal domain of interest. In this work, we extend existing oceanographic sampling methodologies to perform Lagrangian observation studies to sample within the context of an advecting feature of interest. We use GPS-tracked Lagrangian drifters to tag a patch of interest, and utilize its periodic position updates to make an AUV perform surveys around it as it gets advected by ocean currents. Two approaches are described and tested in two field trials in 2010 -a one day experiment in June, followed by a five-day offshore experiment in September. Results from the experiments are presented along with the analysis of the sources of error.
Environmental systems are complicated. They include very intricate spatio-temporal processes, interacting on a wide variety of scales. There is increasingly vast amounts of data for such processes from geographical information systems, remote sensing platforms, monitoring networks, and computer models. In addition, often there is a great variety of scientific knowledge available for such systems, from partial differential equations based on first principles to panel surveys. It is argued that it is not generally adequate to consider such processes from a joint perspective. Instead, the processes often must be considered as a coherently linked system of conditional models. This paper provides a brief overview of hierarchical approaches applied to environmental processes. The key elements of such models can be considered in three general stages, the data stage, process stage, and parameter stage. In each stage, complicated dependence structure is mitigated by conditioning. For example, the data stage can incorporate measurement errors as well as multiple datasets with varying supports. The process and parameter stages can allow spatial and spatio-temporal processes as well as the direct inclusion of scientific knowledge. The paper concludes with a discussion of some outstanding problems in hierarchical modelling of environmental systems, including the need for new collaboration approaches.
The narrow continental shelf of the Southern California Bight (SCB) is characterized by elevated primary productivity relative to the adjacent open ocean. This persistent gradient is maintained by the nitrate fluxes associated with internal waves of tidal frequency (the internal tide). Here we provide the first estimates of the internal-tide–driven horizontal fluxes of nitrate, heat, energy, and salinity, calculated from high-resolution, full water-column data gathered by an autonomous wave-powered profiler and a bottom-mounted current meter. The vertically integrated nitrate, heat, and energy fluxes were onshore over the 3-week period of the field experiment. The inner-shelf area- and time-averaged dissipation rate due to the onshore horizontal energy flux, 2.25 × 10 − 7 W kg − 1, was elevated relative to open ocean values. The magnitude of the vertically integrated horizontal nitrate flux (136.4 g N m − 1 d¹) was similar to phytoplanktonic nitrate uptake rates over the inner-shelf. This nitrate flux was variable in time, capable of supporting 0–2800 mg C m − 2 d − 1 (mean approx. 774 mg C m − 2 d − 1) of “new” primary productivity, depending on the energetics of the internal tide and the cross-shore distribution of nitrate. We postulate that the horizontal, internal-tide–driven nitrate flux is the primary cause of the persistently elevated phytoplankton biomass and productivity over the narrow SCB inner shelf. Furthermore, these results suggest that horizontal fluxes of nutrients driven by internal waves may contribute significantly to primary productivity along the boundaries of aquatic environments.
Chlorophyll concentration, phytoplankton biomass, and total and nitrate-fueled primary productivity increase toward the coast over the 12-km-wide continental shelf of the southern portion of the Southern California Bight. These gradients are accompanied by changes in phytoplankton community composition: the outer shelf is characterized by offshore assemblages including pelagophytes and oligotrophic Synechococcus ecotypes while the inner shelf is dominated by diatoms, coastal Synechococcus ecotypes, and the picoeukaryote Ostreococcus. Across the small horizontal scale of the shelf, large changes in the vertical distribution and flux of nitrate maintain elevated productivity, driving variability in the vertical distribution of biomass and the integrated biomass and productivity of the entire shelf. Temporal variability from hours to days in chlorophyll fluorescence as measured by an autonomous profiling vehicle demonstrates that phytoplankton respond vigorously and rapidly to physical variability. The interaction of physical processes at different temporal and spatial scales is responsible for the observed biological gradients. These dynamics include: (1) vertical shear in the alongshore currents, (2) local wind forcing, (3) the internal tide, and (4) remote, large-scale variability. Individually, these mechanisms rarely or never explain the phytoplankton community composition and metabolic rate gradients. These results and a reanalysis of historical data suggest that monitoring thermal stratification at the shelf break and the tilt of the thermocline across the shelf will augment our ability to predict phytoplankton productivity, community composition, and biomass, thereby improving our understanding of fisheries dynamics and carbon cycling in the coastal ocean.