Comparison of a new qualifier method for multiple object tracking in robocup 2D simulation league
Chapter in Scopus
- Additional Document Info
- View All
© Springer International Publishing Switzerland 2014.In this document two methods for a multiple object tracking problem are tested and compared in a 2D environment with quantisied vision considering the tracking problem as a constraint satisfaction problem as a general approach. The first method is a qualifier method which uses three probabilistic models (identity, distance, and movement direction) to compute the belief of the path of a given object considering the path a Markov process. The second method are particle filters with penalised predictions which expands the belief of a given objects in order to get the best match for it. Each method was tested in two situations. In the first situation the observer was static in a fixed position while the second situation involves a dynamic observer. The methods obtained an almost perfect result of 98% of correct matches for the first situation and achieved a result of nearly 78% of correct matches in the second situation.