Method for the computer-aided calculation of the movement of an object using sensor data

A method for a computer-aided calculation of a movement of an object using sensor data of a sensor disposed on the object, wherein the sensor data contain measurement point sets acquired at different points in time and measurement points from an environment of the object, an object movement between a first measurement point set acquired at a first point in time and a second measurement point set acquired at a second point in time being determined as follows:

• a) extracting structural elements of one or more structure types from the first measurement point set and the second measurement point set using a computer resulting in
• a first data set based upon the first measurement point set comprising extracted structural elements, and
• a second data set based upon the second measurement point set comprising extracted structural elements, and
• unassigned measurement points which cannot be assigned to any structural element;
• b) associating the extracted structural elements of a same structure type of the first data set with the second data set using the computer, and performing a first transformation sequence for mapping the extracted structural elements of the second data set onto the structural elements of the first data set based upon the same structure type using the computer;
• c) associating the extracted structural elements of the same structure type of the first transformation sequence with the unassigned measurement points using the computer, and performing a second transformation sequence for mapping the extracted structural elements of the same structure type of the first transformation sequence onto the unassigned measurement points using the computer;
• and
• d) determining a movement of the object between the first point in time and the second point in time based upon the first transformation sequence and the second transformation sequence,
• wherein in steps b) and c) a measurement noise of the sensor is taken into account, and wherein the first transformation sequence of step b) is determined by an iteration, comprising:
• b.i) for a set of structural element pairs, each of which comprises a structural element of a transformed second data set and a structural element of the first data set, calculating Mahalanobis distances of the transformations, each of which maps the structural elements of a structural element pair onto one another, the Mahalanobis distances being dependent on an instantaneous covariance matrix,
• selecting a structural element pair based upon the Mahalanobis distances, and
• performing a transformation for mapping the structural elements of the selected structural pair onto one another, wherein a new, transformed second data set is obtained as a result;
• and
• b.ii) deleting at least the selected structural element pair from the set of structural element pairs,
• updating the covariance matrix, and
• returning to step b.i),
• wherein in a selection of the transformations in step b.i), transformations with small Mahalanobis distances and/or transformations between long structural elements and/or accumulations of similar transformations are preferred, and
• wherein the transformation performed in step b.i) is estimated based on a Kalman Filter, and/or in step b.ii) the covariance matrix is updated with a Kalman filter.