Planning a collision avoidance model for ship using genetic algorithm

Using Genetic Algorithms to plan the safe path for ship in congested traffic situation, a new gene vector is proposed. The gene vector is composed of position and speed of own ship, as well as a noise model. The noise model describes the influence to maneuvering ships system resulting from wind, sea wave and the other natural factors. To test and verify the new gene vector, the equipments installed on "Shioji Maru" that is the training ship of our university, have been applied to build an automatic collision avoidance system. In the experimental system the ARPA (Automatic Radar Plotting Aids) system was used to collect the information of navigational obstacles around own ship. The information was processed. The useful information especially related to target ships was extracted and used to derive a stochastic predictor that can predict the future position and degree of future collision threat in enough future time. The own information was detected with GPS and the other sensors. These data were introduced to GA optimum controller. The optimum or semi-optimum path was evolved from a set of possible safe paths based on the fitness function. A lot of experiments have been done and the experimental results have been presented.