Energy efficient navigational methods for autonomous underwater gliders in surface denied regions

摘要

Autonomous underwater gliders routinely perform long duration profiling missionsudwhile characterizing the chemical, physical and biological properties of the waterudcolumn. These measurements have opened up new ways of observing the ocean’s processesudand their interactions with the atmosphere across time and length scales whichudwere not previously possible. Extending these observations to ice-covered regions is ofudimportance due to their role in ocean circulation patterns, increased economic interestudin these areas and a general sparsity of observations.udThis thesis develops an energy optimal depth controller, a terrain aided navigationudmethod and a magnetic measurement method for an autonomous underwater glider.udA review of existing methods suitable for navigation in underwater environments asudwell as the state of the art in magnetic measurement and calibration techniques isudalso presented.udThe energy optimal depth controller is developed and implemented based on anudintegral state feedback controller. A second order linear time invariant system isudidentified from field data and used to compute the state feedback controller gainsudthrough an augmented linear quadratic regulator. The resulting gains and stateudfeedback controller methodology are verified through field trials and found to controludthe depth of the vehicle while losing less than one percent of the vehicle’s propulsiveudload to control inputs or lift induced drag.udThe terrain aided navigation method is developed based on a jittered bootstrapudalgorithm which is a type of particle filter that makes use of the vehicle’s deadreckonedudnavigation solution, onboard altimeter and a local digital parameter modelud(DPM). An evaluation is performed through post-processing offline location estimatesudfrom field trials which took place in Holyrood Arm, Newfoundland, overlapping audpreviously collected DPM. During the post-processing of these trials, the number ofudparticles, jittering variance and DPM grid cell size were varied. Online open loopudfield trials were performed through integrating a new single board computer. In theseudtrials the localization error remained bounded and improved on the dead reckoninguderror, validating the filter despite the large dead-reckoned errors, single beam altitudeudmeasurements, and short test duration.udTerrain aided navigation methods perform poorly in regions of flat terrain orudin deep water where the seafloor is beyond the range of the altimeter. Magneticudmeasurements of the Earth’s main field have been proposed previously to augmentudterrain aided navigation algorithms in these regions. To this end a low power magneticudinstrumentation suite for an underwater glider has been developed. Two calibrationudmethodologies were also developed and compared against regional digital models ofudthe magnetic field. The calibration methods include one for which the actuators in theudvehicle were kept in fixed locations and a second for which the calibration coefficientsudwere parameterized for the actuator locations. Both methods were found to agreeudwith the low frequency content in the a-priori regional magnetic anomaly grids.