Development of Object Tracking System Using Remotely Operated Vehicle Based on Visual sensor

摘要

The use of the camera on ROV aims to visualize the ROV work environment so as to facilitate the operator on the surface to control the movement of ROV. In an attempt to track an object, it is too difficult for the operator to maintain object position that remains in point of view camera. The number of disturbances such as water currents can cause the object to disappear from point of view camera. For that reason, we developed an object tracking system on ROV (Remotely Operated Vehicle) that can approach object autonomously by utilizing camera as a visual sensor. The camera on the ROV is utilized as a visual sensor that sends feedback in the actual position of the ROV to determine the direction movement against the object. Tracking method is applied by utilizing the color as a feature because this method is very good in detecting objects with stable lighting. Real-time captured images will be processed with image processing to obtain a relative estimate of the object position. The estimate value of the object position is sent to the controller to move the ROV according to the direction and position of the object. Autonomous object tracking system on the ROV is divided into several stages; the first stage is image processing to detect the shape and the actual position of the object. The next stage is to connect the whole system. The performance of the developed system during the experiment takes about 20 seconds to reach ideal position. The image generated by image processing will be described in this paper.