Model based design of a stereo vision system for intelligent deep-sea operations

Abstract

A stereo vision system for deep-sea operations is presented. The system consists of cameras in pressure bottles, which are daisy-chained to a computer bottle. The system has substantial computation power for on-board stereo processing as well as for further computer vision methods to support autonomous intelligent functions, e.g., object recognition, navigation, mapping, inspection, and intervention. The model based design presented here includes two main aspects. First, a formalized approach to the component selection for the stereo set-up is introduced, i.e., given especially accuracy and baseline constraints as well as lens and imager options, an algorithmic analysis is provided. This approach is also of interest for the design of stereo systems in general. Second, the specific aspects of deep sea operations are addressed. This includes especially the validation and optimization of the pressure bottles for the cameras with a Finite Element Method (FEM). Experiments and results are presented, which include a validation of the stereo performance in air, robustness tests of the bottles in pressure tanks, and field trials of the complete system off the shore of Marseille on a commercial Remotely Operated Vehicle (ROV).