Part of special issue: 13th IFAC Conference on Control Applications in Marine Systems, Robotics, and Vehicles
Automatic Identification System (AIS) increasingly serves as the primary sensor for decision making. But the provided information can be manipulated, and the signal can even be manually disabled to hide illegal activities. In addition, the quality of the AIS data depends strongly on the quality of the sensors installed on the vessel. To determine the accuracy of AIS position data, the paper will compare AIS data with associated radar tracks. The radar data is assumed to be ground truth so that the accuracy of AIS can be determined. In addition, various existing AIS interpolation approaches are discussed and executed on a common representative dataset with over 6.2 million data points of the German Bight. This allows to make meaningful statements about the quality of the methods when interpolating AIS data. Overall, it was revealed that the mean discrepancy between AIS and radar data measures about 97.72m. Moreover, some presented approaches can approximate AIS data with higher accuracy than the classical linear interpolation. However, this accuracy is also associated with an increased computation time, so that it should be weighed which interpolation method should be used depending on the use case.