2012 · 14 citations · 6 references
EngineeringAutonomous SystemsRobotics EngineeringAutonomous VehiclesMarine RoboticsRoboticsAutonomous NavigationNavigation SystemsObstacle AvoidanceRobot NavigationSense-and-avoid SystemsVisual NavigationReactive AvoidanceReactive Avoidance ControlReactive Obstacle AvoidanceRelative NavigationTerrain-referenced NavigationTerrain-relative NavigationUnmanned Aerial Vehicle NavigationVision-based Navigation
A motivating mission for AUVs is to collect a time series of images of a benthic site to monitor it for change. This mission includes performing a visual survey of an area of the seafloor and then returning to selected sites within that survey area on subsequent visits. To enable this capability for remote sites far from the launch point, an AUV designed for long-distance travel is required. Such AUVs are typically motion-constrained - they cannot hover and must maintain forward flight for controllability. In addition to a navigational system capable of returning the vehicle to the site, a terrain-following system is required to allow the motion-constrained AUV to fly safely within a few meters of the seafloor to collect images. Recent demonstrations using MBARI's Doradoclass AUVs combined with a Terrain-Relative Navigation system (TRN) have proven much of the navigational capability, demonstrating return-to-site within approximately 3 m. Imaging of the seafloor using these AUVs has also been demonstrated using a reactive obstacle avoidance control law. While successful, this reactive-only system is conservative, resulting in sections of the seafloor being missed during the imaging process. This paper presents an approach for planning terrain-following trajectories for an AUV that will allow it to operate safely in close proximity to rugged terrain. The approach fuses reactive obstacle avoidance with anticipatory information from the TRN system. Specifically, by including knowledge of known terrain ahead, a more aggressive trajectory can be planned, resulting in improved mission performance without compromising vehicle safety. A reactive system is still incorporated, but only to handle any unmapped obstacles that are encountered. The new terrain-following algorithm is described, and its feasibility is demonstrated through simulations using field data from AUV operations in Monterey Bay.
6
Path Planning for Autonomous Underwater Vehicles
Clment Petres, Yan Pailhas, Pedro Patrón et al. · IEEE Transactions on Robotics · 2007 · 430 citations