Draft: Modeling and Inspection Applications of a Coastal Distributed Autonomous Sensor Network

Real time in-situ measurements are essential for monitoring and understanding physical and biochemical changes within ocean environments. Phenomena of interest usually display spatial and temporal dynamics that span different scales. As a result, a combination of different vehicles, sensors, and advanced control algorithms are required in oceanographic monitoring systems. In this study our group presents the design of a distributed heterogeneous autonomous sensor network that combines underwater, surface, and aerial robotic vehicles along with advanced sensor payloads, planning algorithms and learning principles to successfully operate across the scales and constraints found in coastal environments. Examples where the robotic sensor network is used to localize algal blooms and collect modeling data in the coastal regions of the island nation of Singapore and to construct 3D models of marine structures for inspection and harbor navigation are presented. The system was successfully tested in seawater environments around Singapore where the water current is around 1-2m/s. ∗Address all correspondence to this author. INTRODUCTION This study presents the autonomous sensor network being developed at the Center for Environmental Sensing and Modeling (CENSAM) in Singapore and its applications to coastal environments. Ocean observing and prediction systems present challenges for vehicles, sensors, motion planners, data assimilation and predictive models. In addition, coastal environments have particular challenges such as low depths and commercial vehicle traffic which increase the likelihood of collisions. Our group is developing a distributed heterogenous autonomous sensor network that combines underwater, surface, and aerial robotic vehicles along with advanced sensor payloads, planing algorithms and learning principles to successfully operate across the scales and constraints found in coastal environments. Its effectiveness relies on the seamless operation of all vehicles, their safe interaction with each other and the environement, the capability of collecting pertinent data, and the capabity to analyze the collected data and use information within to optimze the sampling process. The two applications explored herein portray examples of basic tools for these objectives. The study first presents a description of the configuration and components of the autonomous sensor network developed by our group and subsequently two applications are described in detail: a procedure for algal bloom monitoring and 3D surface reconstruction. Preliminary results are 1 Copyright c © 2012 by ASME summarized within each section and conlcuding remarks and future work are outlined. AUTONOMOUS SENSOR NETWORK Ocean observing and prediction systems have inspired both theoretical and experimental work. The Autonomous Ocean Sampling Network (AOSN) project [1, 2] outlined the significant challenges for the practical implementation of such systems. Other studies since have explored both hardware (vehicles and sensors) [3, 4] and software (motion planners, data assimilation, predictive models) [5–7] for use in robotic sensor networks to observe and forecast conditions in marine environments. This study presents an update on the autonomous sensor network being developed at CENSAM [8] and its applications to coastal environments. Fig. 1 shows a diagram of the typical spatial distribution of the active network nodes within CENSAM’s network, which include research vessels, autonomous surface vehicles (ASVs), autonomous underwater vehicles (AUVs), and unmanned aerial vehicles (UAVs). Large distances (∼ 103 meters) are covered by the air vehicles (quadrotors) whose mission is to rapidly scan areas of interest and identify potential target zones. In addition to navigation related hardware, quadrotor sensor payloads include digital and infrared cameras. If a quadrotor identifies a feature of interest its location is geo-tagged and relayed to the rest of the network, subsequently a research vessel deploys autonomous underwater vehicles (Iver2 submarines from Oceanserver) and autonomous surface vehicles (SCOUT ocean kayaks equipped for oceanographic and undersea testing) within the vicinity of the areas of interest (∼ 102 meters). The AUVs and ASVs perform finer scanning passes inside the target zones to provide denser data for further analysis. In addition to inertial navigation and communication equipment, the surface and underwater vehicles are equipped with oceanographic sensors to measure pH, temperature, conductivity, chlorophyll, rhodamine, turbidity, and dissolved oxygen. Spectrophotometers (AC-9 from Wetlabs) are also used to measure the absorption and attenuation coefficients of the elements within the sampled zones. ALGAL BLOOMS The first application discussed for the robotic sensor network is the location and scientific survey of algal blooms in the coastal regions of Singapore. Harmful algal blooms are an increasing problem in the coastal waterways surrounding Singapore due to urban development as well as longer term climate changes. A toxic algal bloom in December 2009 resulted in 200,000 fish killed in the Pasir Ris area alone, and a second bloom by a different algal species occurred a month later in the same location. An understanding of the underlying ecological, chemical, tidal, and hydrodynamic factors is needed to develop reliable models of these events. However, modeling a system FIGURE 1. Stratified heterogeneous sensor network. Aerial, surface and underwater vehicles along with research vessels are used to span different spatial and temporal scales and optimally scan and monitor