Extensive work has been done in the field of autonomous robotics in recent years. Most of this work has been done on ground vehicles, with some work carried out on air vehicles and underwater vehicles. The area of autonomous surface vehicles (ASVs) has been largely unexplored, and this is why the Maritime Competitions were created. These competitions are mission based, and task an autonomous surface vehicle with the negotiation of many realistic challenges. These tasks include channel navigation, obstacle detection and avoidance, docking, acoustic positioning and observation.
There are many challenges inherently present in the autonomous control of surface vehicles for these competitions. As GPS sensors are not very precise, most of navigation is determined by the vision system. The competitions are held outdoors, which means that the lighting conditions vary often, and tend to oversaturate the image. This oversaturation makes color detection very difficult, as even red buoy are detected as almost entirely white. The vision system also has to accommodate for reflections that can distort the image of objects on the surface, and objects in the background that could be confused for targets (such as a tree that could be detected similarly to a green buoy). Some targets vary in size, shape and color, which means the vision system must be able to detect depth to locate the obstacles; in other words, distance cannot be inferred by the size in the image. There are also issues present for other sensors, such as the increased speed at which sound travels through water - which makes acoustic positioning more difficult than it would be above the surface.
The main focus of our approach is on vision based sensing and analysis for navigation.
We have looked at various issues relating to this problem.