Michael P. Hamilton

Experiments with Underwater Robot Localization and Tracking

This paper describes a novel experiment in which two very different methods of underwater robot localization are compared. The first method is based on a geometric approach in which a mobile node moves within a field of static nodes, and all nodes are capable of estimating the range to their neighbours acoustically. The second method uses visual odometry, from stereo cameras, by integrating scaled optical flow. The fundamental algorithmic principles of each localization technique is described. We also present experimental results comparing acoustic localization with GPS for surface operation, and a comparison of acoustic and visual methods for underwater operation.

Use of a Networked Digital Camera to Estimate Net CO2 Uptake of a Desiccation-Tolerant Moss

Simple visible‐light digital cameras offer a potential for expanded forms of plant ecological research. The moss Tortula princeps undergoes changes in reflected visible light during cycles of drying and hydrating in the field, and the MossCam project has collected digital images of T. princeps at least daily since 2003. Laboratory studies can be used to calibrate these images to indicate field physiological conditions. Drying the moss 6 d in the laboratory resulted in a decrease of net CO2 uptake to near 0; recovery after rewetting occurred within 10 min. The difference in reflectance between hydrated and dry T. princeps was maximal ca. 550 nm, and maximal net CO2 uptake was linearly related to the green:red ratio of laboratory images when net CO2 uptake was positive. Using the green:red ratio of field images and otherwise assuming ideal conditions, the total carbon gain for a 6‐d period around a 1.3‐mm rain event was ca. 208 mmol CO2 m−2, equivalent to 69 d of respiration under dry conditions. Using a visible‐light digital camera with micrometeorological data and laboratory‐based gas exchange measurements, T. princeps can be used as a model species for simple field estimations of photosynthesis, carbon gain, and phenological events.

Heartbeat of a nest: Using imagers as biological sensors

We present a scalable end-to-end system for vision-based monitoring of natural environments, and illustrate its use for the analysis of avian nesting cycles. Our system enables automated analysis of thousands of images, where manual processing would be infeasible. We automate the analysis of raw imaging data using statistics that are tailored to the task of interest. These “features” are a representation to be fed to classifiers that exploit spatial and temporal consistencies. Our testbed can detect the presence or absence of a bird with an accuracy of 82%, count eggs with an accuracy of 84%, and detect the inception of the nesting stage within a day. Our results demonstrate the challenges and potential benefits of using imagers as biological sensors. An exploration of system performance under varying image resolution and frame rate suggest that an in situ adaptive vision system is technically feasible.

Perspectives on next‐generation technology for environmental sensor networks

Sensor networks promise to transform and expand environmental science. However, many technological difficulties must be overcome to achieve this potential. Partnerships of ecologists with computer scientists and engineers are critical in meeting these challenges. Technological issues include promoting innovation in new sensor design, incorporating power optimization schemes, integrating appropriate communication protocols, streamlining data management and access, using innovative graphic and statistical analyses, and enabling both scientists and the public to access the results. Multidisciplinary partnerships are making major contributions to technological advances, and we showcase examples of this exciting new technology, as well as new approaches for training researchers to make effective use of emerging tools.

Geographic information systems: Providing information for wildland fire planning

Controlling wildfires within the wildland/urban interface has proven to be the most complex challenge facing wildland fire agencies. Although program improvements to increase the efficiency of interface suppression efforts have been suggested, the availability of information about the wildfire environment remains a critical resource for wildland fire planning. Geographic Information Systems (GISs) can provide the technology to store, manipulate, analyze, and display spatially oriented information in a form necessary for efficient fire planning and incident decision making. Complex map and attribute information, including vegetation types, fuels models, weather patterns, topography, fire suppression environment, landuse characteristics, and microenvironmental features, can be rapidly summarized and integrated. This integrated information can be used to create unique polygons useful in predicting fire behavior, allocating fire suppression resources, and as an aid in planning land use. Simplified user interfaces and the portability of new hardware systems will allow GISs to be used at every level of wildland fire planning.

EcoIP: An open source image analysis toolkit to identify different stages of plant phenology for multiple species with pan-tilt-zoom cameras

Because of the increased number of cameras employed in environmental sensing and the tremendous image output they produce, we have created a flexible, open-source software solution called EcoIP to help automatically determine different phenophases for different species from digital image sequences. Onset and ending dates are calculated through an iterative process: (1) training images are chosen and areas of interest identified, (2) separation of foreground and background is accomplished based on a naive Bayesian method, (3) a signal is created based on the separation model and (4) it is then fit to a sigmoid that contains the dates of interest. Results using different phenological events of different species indicate that estimated dates fall within a few days of the observed dates for most cases. Our experiments indicate that color separability and scene illumination are contributing factors to this error. EcoIP is implemented as an open platform that encourages anyone to execute, copy, distribute, study, change, and/or improve the application.

Scientific data visualization and biological diversity: new tools for spatializing multimedia observations of species and ecosystems

Abstract Current knowledge of the biodiversity of protected ecosystems is often limited to museum collection specimen data, outdated or cursory inventories and anecdotal accounts. Beyond the inventory of biological diversity comes the need to monitor changes of many parameters and at many scales, and the need to incorporate this knowledge into an accessible information system for biodiversity management planning and conservation education. This paper briefly describes our continuing research efforts to develop new ways to collect, analyze, organize and distribute biodiversity data.