A Short Study of Statistics of Collisions Between Merchant Vessels

Last year the Sperry Piedmont Company was awarded a contract by the Maritime Administration to determine the feasibility of a Lookout Assist Device. The study has involved virtually all important aspects of the problem from physics to economics. It has long been recognized that a need exists for equipment to aid the ship's lookout in his task of detecting vessels and other objects in the sea that represent potential colission hazards. On the other hand, the prudent ship operator will probably not purchase equipment to aid the lookout in the prevention of collision unless it can be demonstrated that its purchase will save him money in the logn run. Many practical questions must be answered before the design goals of a lookout assist system can be defined. For instance, at what rantge should the device warn of an intruding vessel? What bearings are most threatening? What is the dollar value of a lookout assist device? Much good work has been done by others in the collection and assimilation of collision statistics that have helped us in answering these and similar questions. Use has been made of the computer or probability theory when the statistical data was meagre or lackign altogether. Occasionally the investigator found vagueness or ambiguity in terminology. When this happened, he arbitrarily stated the problems in newly defiend terms, hopefullly not too far afield from traditional parlance, in order to obtain any answers at all. Unfortunately, the investigator quickly discovered that the best he could hope for was an order of magnitude in the accuracy of the results. The reasons for this are obvious when one condiers the difficuties in obtaining large data samples and the hazardous task of drawing reliable inferences from the data. Nevertheless, progress in collision prevention could not wait for the accumulation of collision data and it was necessary to do our best with what was available.