Visual servo control of an underwater vehicle using a laser vision system

This paper describes a position-based visual servo control scheme designed for an underwater vehicle. The methodology proposes a path planning technique, which guarantees that a flat target is kept in the camera optical field, while the vehicle avoids collision with the surface the target lays on. The vehicle pose (position and orientation) with respect to the target is obtained using a Laser Vision System (LVS). The LVS projects two laser dots in the image plane while it tracks the target using computer vision algorithms. The position of each laser dot in the image plane is directly related to the distance between the vehicle and the surface the target is located. The path planning strategy is based on the Artificial Potential Field method (APF). The attractive part of the APF is responsible for minimizing the error between the current vehicle position and the desired. The repulsive part of the APF restricts the target inside the camera optical field while keeps away the laser dots from image regions related to small distances between the vehicle and the surface the target is located. The steering control of the vehicle is achieved by feeding the computed points of the path planning into a Cartesian kinematic controller, which was slightly modified for the needs of the methodology. The overall efficiency of the system, was proved through an extensive experimental procedure, using a small Remotely Operated Vehicle (ROV) in a test tank.