Progress towards underwater 3D scene recovery

摘要

The underwater environment presents many challenges for robotic systems and sensors. Not only is it difficult to determine appropriate locomotive and control strategies, sensing underwater is plagued by highly variable lighting, dynamic objects, and suspended particulate matter. Despite these challenges the aquatic environment presents many real and practical applications for autonomous robots. Fundamentally, these tasks require knowledge of the 3D environment, the robot's location within the environment, and reactive vehicle control. In this paper we describe solutions to the problem of providing effective control of underwater robotic systems that can be used to obtain accurate models of underwater structures. Two specific components of the research are described here: (i) a 3D stereo-vision sensor that integrates stereo vision imagery with inertial measurements, and (ii) a tablet-based control interface that can be used to control the process of data collection.