Vision based control of autonomous underwater vehicles

Vision based control (Visual Servoing) of an autonomous 6-DOF vehicle is considered in this thesis. The vehicle is to be navigated underwater using the cameras mounted on its body. With the new developed algorithm, the need for underwater GPS, Tilt Sensors and other costly devices is eliminated. The new algorithm overcomes the shortcomings of the existing algorithms, PBVS (Position-Based Visual Servoing) and IBVS (Image-Based Visual Servoing). The underwater vehicle is navigated using the image feedback obtained from the cameras. These image features are fed to the control system on the vehicle which in turn generates the desired motion. Simulations runs in MATLAB/SIMULINK give satisfactory results of the algorithm.