Logo des Repositoriums

Auflistung nach Autor:in "Kester, Leon"

1 - 2 von 2
  • Konferenzbeitrag
    Adaptive state multiple-hypothesis tracking
    (INFORMATIK 2006 – Informatik für Menschen, Band 1, 2006) Kleef, Jelle van; Kester, Leon
    In tracking algorithms where measurements from various sensors are combined the track state representation is usually dependent on the type of sensor information that is received. When a multi-hypothesis tracking algorithm is used the probabilities of the different hypotheses containing tracks in different representations need to be re-evaluated when track state representations are changed. For the particular case of trilateration a method is presented to adapt the state representation as more information becomes available. A discussion is given on how to re-evaluate the probabilities of the hypotheses leading to a method for the trilateration case. This is illustrated by a simple example.
  • Konferenzbeitrag
    An improved method for creating shared belief in communication constrained sensor networks
    (Informatik 2009 – Im Focus das Leben, 2009) Foeken, Eelke van; Hiemstra, Peter; Kester, Leon
    Many application domains, such as public safety, traffic management, crisis management and defense are examples where networked systems are used for gathering sensor data. In general these networked systems are faced with a growing need for adaptive behavior in observing and acting in dynamically changing environments. Designing a system which is adaptive to dynamically changing environments becomes dif- ficult, particularly a decentralized networked system that is appropriate for multi-platform situation awareness. We have chosen to design the sensor network following the Networked Adaptive Interactive Hybrid Systems (NAIHS) model [1]. Using this model the data processing chain is decomposed into functional components. These functional com- ponents interact by requesting information they need and fulfilling requests received from other components. Local evaluation of the available data with respect to the different requests and available resources is a key process in each component. An improved eval- uation method is presented which is capable of locally balancing the information value against the resource costs of data. The experiments show that the evaluation method with the same amount of communication costs, results in a higher common picture quality with multiple objects and overlapping detection ranges, in comparison to the benchmark method.