UC Santa Cruz

Autonomous surface vehicles (ASVs) may be one of the most promising tools to help study, protect, and address problems within complex ocean ecosystems. ASVs can be used to collect ocean data for climate modeling, collect plastics and other human-related pollution, study ocean fronts, and map the ocean floor. These types of vehicles possess sensors that can be used with spatial estimation techniques, such as simultaneous localization and mapping (SLAM), ordinary kriging, or Gaussian process regression (GPR). Such algorithms produce an estimate of the field representing the local environment with an associated level of uncertainty. As the ASV collects more measurements, the estimate is updated. However, it is common for ASVs to follow static paths with linear waypoint-to-waypoint tracking for navigating a field of interest; the path doesn't change as more information of the field is obtained. An alternative strategy is dynamic path planning based on uncertainty suppression. In this thesis an optimal exploration algorithm is presented that maximizes the current field-estimate variance along a path. This algorithm is based on Bellman-Ford graph search and is compared to a myopic, zigzag, and other planners in simulation. These path planners are also paired with different spatial estimation methods, including partitioned ordinary kriging (POK) and partitioned GPR. The results are discussed including the computation time trade-offs. The intent is to determine an efficient path planner and spatial estimator that can be used at run-time onboard an ASV.

Two experimentally implemented ASVs, the Slug 2 and Slug 3 are discussed. This includes discussion of their system design, attitude and heading reference system, system modeling, system identification, control, and experimental results. Both ASVs used GPS for waypoint tracking and data logging. The Slug 3 used an echo sounder depth sensor to measure the depth of a small body of water. For navigation the ASVs used a GPS module and an IMU with tri-axis MEMS accelerometers, gyroscopes, and magnetometers.