Achieving real-time Intelligent performance in underwater robotics using cheap off-the-shelf components

The present paper proposes a new simple strategy to build underwater robots using cheap off-the-shelf components. The proposed strategy successfully achieves real-time operation without requiring expensive special purpose hardware/software. The simple geometry of the proposed design makes it possible to derive an accurate dynamic model of the robot. The dynamic model, hardware/software preprocessing of webcam images, multi-feature Kalman based combining as well as neural networks are used to compensate for the limited sensors and computational resources.