Autonomous systems like inland vessels require knowing the behavior of surrounding vessels and moving objects. Predicting the behavior of surrounding inland vehicles operating in a narrow field like rivers, channels, etc. around the Ego-system is challenging due to the required accuracy. Existing approaches for sea navigation cannot be used because the precision requirements are lower than required for inland vessels navigation. Precise behavior prediction is required that allows navigation with high precision during overtaking in upstream and downstream directions. In this contribution, new approaches have been developed using past trajectories information of different or similar types of inland vessels. Here the concepts of three approaches to predict the behavior, based on AIS data, are discussed and compared. In the first approach, predictions are done with a model developed using simple parameter-based approach. The predictions are based on the global model parameters of the vessel and local adaption. In the second approach, the Bayesian approach is applied to define the best trajectory (intention) from the clustered past information. In the third approach, the two approaches are combined; here, the local parameters from the first model and the intentions from the second model are taken into account so that the prediction errors are reduced. The initial results, from this study, are based on the data of a single ship. A further extension of the approach will consider data from several vessels of the same type.
Probabilistic ship behavior prediction using generic models
2022-10-08
1954997 byte
Conference paper
Electronic Resource
English
Probabilistic Fatigue Life Prediction for Ship Structures Using Fracture Mechanics
British Library Conference Proceedings | 2000
|Probabilistic Models for load effects in ship structures
TIBKAT | 1984
|Impact scenario models for probabilistic risk-based design for ship-ship collision
British Library Online Contents | 2013
|