Devices, algorithms and systems for maritime persistent surveillance

This thesis presents a novel approach to long-term marine data collection and monitoring. Long-term marine data collection is a key component for understanding planetary scale physical processes and for studying and understanding marine life. Marine monitoring is an important activity for border protection, port security and offshore oil field operations. However, monitoring is not easy because salt water is a harsh environment for humans and for instruments. Radio communication and remote sensing are difficult below ocean surface. Our approach to ocean data collection relies on the integration of (1) a network of underwater sensor nodes with acoustic and optical communication, (2) an autonomous underwater vehicle (AUV) and (3) a novel sensing device. A key characteristic is the extensive use of visible light for information transfer underwater. We use light for sensing, communication and control. We envision a system composed of sensor nodes that are deployed at static locations for data collection. Using acoustic signaling and pairwise ranging the sensor nodes can compute their positions (self-localize) and track mobile objects (e.g., AUVs). The AUV can visit the sensor nodes periodically and download their data using the high speed, low power optical communication. One consequence of using optical communication for the bulk of the data transfer is that less data needs to be transferred over the acoustic links, thus enabling the use of low power, low data rate techniques. For navigation, the AUV can rely on the tracking information provided by the sensor network. In addition, the AUV can dock and transport sensor nodes efficiently, enabling their autonomous relocation and recovery. The main application of our system is coral reef ecosystem research and health monitoring. In this application the robot and the sensor nodes can be fitted with our novel imaging sensor, capable of taking underwater color-accurate photographs for reef health assessment and species identification. Compared to existing techniques, our approach: (1) simplifies the deployment of sensors through sensor self-localization, (2) provides sensor status information and thus enables the user to capture rare events or to react to sensor failure, (3) provides the user real time data and thus enables adaptive sampling, (4) simplifies mobile sensing underwater by providing position information to underwater robots, (5) collects new types of data (accurate color images) through the use of new sensors. We present several innovations that enable our approach: (1) an adaptive illumination approach to underwater imaging, (2) an underwater optical communication system using green light, (3) a low power modulation and medium access protocol for underwater acoustic telemetry, (4) a new AUV design capable of hovering and of efficiently transporting dynamic payloads. We present the design, fabrication and evaluation of a hardware platform to validate our approach. Our platform includes: (1) AQUAN ET, a wireless underwater sensor network composed of AQUAN ODES, (2) AMOUR, an underwater vehicle capable of autonomous navigation, data muling, docking and efficient transport of dynamic payloads and (3) AQUALIGHT an underwater variable-spectrum Xenon strobe which enables underwater color accurate photography. We use this platform to implement and experimentally evaluate our algorithms and protocols. (Copies available exclusively from MIT Libraries, Rm. 14-0551, Cambridge, MA 02139-4307. Ph. 617-253-5668; Fax 617-253-1690.)