Mapping for control in an underwater environment using a dynamic inverse-sonar model

摘要

Autonomous Underwater Vehicles (AUVs) are commonly used in oceanographic applications such as seafloor survey and underwater iceberg profiling. During the surveys, the vehicles are intended to follow the variation of the seafloor or iceberg surface at a constant stand-off distance in order to maintain a consistent sensor footprint. Mechanical scanning sonar are usually chosen for measuring the distance from the vehicle to the object. Due to the wide beamwidth of the sonar, the uncertainty from the unknown direction of the received echo will affect the accuracy of the resulting environmental map. On the mobile robots, a static inverse-sonar model is introduced to compensate for such uncertainty for range finders. As an improvement, we present a dynamic inverse-sonar model accounting for the trend of the surveying environment, i. e. the terrain elevation. A vehicle-attached occupancy map is introduced for estimating the terrain elevation, while a global occupancy map is created to present the surrounding environment. The proposed technique is first simulated in mapping a vertical underwater column. The resulting global map is found to be more accurate in presenting the original underwater features compared to the results from a static inverse-sonar model. The technique is further applied to a set of sonar-data collected from an iceberg profiling trial.