Any object or entity moving i n an open environment requires, together with certain navigational skills, the capacity to avoid collisions. Such ability presupposes: manoeuvrability; awareness of the nature of, and ability to detect, potential hazards; and innate intelligence to formulate, evaluate, and implement appropriate avoidance strategies. The maritime environment, and vessels moving within it, provide a classic illustration of the problems and principles of collision avoidance in two-dimensional space. Events in recent years, leading to loss of life, wastage of resources and pollution of the environment, have highlighted the limited success being achieved in this area. In most cases human error has been identified as the major factor; errors of judgement consequent on either a false sense of security or (more often) information overload - an excess of data from various sources, to be processed in limited time. This thesis describes the development of a shipboard expert system for marine collision avoidance. This system, to be operated in a microcomputer on the bridge, analyses all pertinent data on an encounter with another vessel and formulates an optimal strategy to deal with the situation. Detailed information on own-ships characteristics, including a mathematical manoeuvring model, is held in the knowledge base. Situational data relating to own-ship, potential hazards and any other relevant factors are to be input via on-board sensors , including ARPA (Automatic Radar Plotting Aid) ; keyboard input is an additional option . The user interface makes extensive use of an advanced WIMPS (Windows Icons-Mouse-Pointer System) environment; scrollable text and graphic s windows provide the main display; menus and dialogue boxes provide for input and control functions; all user facilities are mouse-operated. The system logic incorporates facilities for emergency manoeuvring, and t h e potential for extension to handle multi-ship encounters; direct input and consideration of electronic chart data on coastal features and shipping lanes has also been anticipated in the design. These issues are considered in some depth in the thesis and form the basis of two new research projects now in progress. The system has been validated for two-ship encounters i n the open sea, through expert appraisal of simulated manoeuvres and random simulation of large numbers of encounters; sea trial s on board a research vessel have shown it to be seaworthy and effective in practical operation.

Document Type


Publication Date