Navigation of an autonomous vehicle in an uncertain environment requires accurate positioning of obstacles, using several sensors such as radar, laser range finder, Forward Looking Infra Red (FUR), day light camera and sonar sensors. The proposed navigation system considers the data about obstacles as its input and is able to navigate toward the target in a safe distance from obstacles with acceptable velocity. Our case study is the navigation of a military boat in the sea. The design is based on a new controller structure, which
is defined for each new detected obstacle. The outputs of these controllers are fused together by a weighted-combination of outputs of existing units. Introduction of these weights and their implementations and learning processes are major characteristics of this research. Other characteristics are its simplicity, fast and robust features with respect to disturbances and noises. Such sensors have been involved during recent years, but the navigation system for vessels and ships (our case study) is still manual for supervising data acquisition and processing. Manual methods always confront some mistakes and in some cases it is needed to guide the vehicle without human aid. In such conditions an autonomous navigator should be installed on the vessel. Such navigator utilizes the data obtained from different sensors in a Data Fusion System (DFS) to estimate the location of obstacles, while dynamic or static and also submarine of floating. Since these information are instantly variable the vessel should also be guided toward the target instantly. Defining a potential function has been the major method used for autonomous guidance and collision avoidance.
)