Logo des Repositoriums

Auflistung nach Autor:in "Ulmke, Martin"

1 - 3 von 3
  • Konferenzbeitrag
    Multi hypothesis tracking of ground moving targets
    (Informatik 2005 – Informatik Live! Band 2, 2005) Koller, Jost; Ulmke, Martin
    We present an application of a multi hypothesis tracking (MHT) algorithm to the case of ground moving targets detected by GMTI (Ground Moving Target Indicator) radar sensors. The algorithms for track extraction, prediction, filtering, and track maintenance are briefly described, and exemplary tracking results for simulated MTI data are demonstrated.
  • Konferenzbeitrag
    Multi-target tracking using signal strength measurements
    (INFORMATIK 2010. Service Science – Neue Perspektiven für die Informatik. Band 2, 2010) Mertens, Michael; Ulmke, Martin
    Many surveillance tasks rely on the generation of stable, continuous tracks of objects of interest. Often, track continuity has a higher priority than track accuracy, as valuable information on the identity or the origin of an object is lost by each track drop. The main source of track fragmentation are missing detections due to a limited field of view or technical or topographical masking. In the multi-target case, in addition, individual tracks can be interchanged by unknown data assignments. Therefore, the exploitation of additional sensor data is required in order to discriminate individual objects and to associate track fragments. As signal strength measurements are standard output of modern radar systems, the integration of this information into a Bayesian tracking scheme is discussed in the present paper. In contrast to previous approaches, the knowledge on the target's signal strength is not only used for an improved calculation of the association probabilities, but it enters into the algorithm as a random variable which is estimated sequentially. By this approach it is not only possible to discriminate closely-spaced targets and improve the track continuity, but also to support possible classification and identification tasks. The signal strength fluctuations of the target returns are modeled by the Swerling-I and Swerling-III cases. As a first performance evaluation, numerical results are presented based on a two-target simulation scenario.
  • Konferenzbeitrag
    Multi-target tracking with partially unresolved measurements
    (INFORMATIK 2010. Service Science – Neue Perspektiven für die Informatik. Band 2, 2010) Svensson, Daniel; Ulmke, Martin; Danielsson, Lars
    The problem of limited sensor resolution, although usually ignored in target tracking, occurs in multi-target scenarios whenever the target distance falls below the size of the sensor resolution cell. Typical examples are the surveillance of aircraft in formation, and convoy tracking for ground surveillance. Ignoring the limited sensor resolution in a tracking system may lead to degraded tracking performance, in particular unwanted track-losses. In this paper, an extension of the resolution model by Koch and van Keuk to the case of arbitrary object numbers is discussed. The model is incorporated into the Joint Probabilistic Data Association Filter (JPDAF) and applied to a simulated scenario with partially unresolved targets.