Comparison of Data Association Methods. Application to Road Obstacle Tracking Using a Doppler Effect Radar
| Autor: | Stéphane Jouannin, J. Gallice, Laurent Trassoudaine |
|---|---|
| Rok vydání: | 1998 |
| Předmět: |
Computer science
business.industry Probabilistic data association filter Tracking system Joint Probabilistic Data Association Filter Simultaneous localization and mapping computer.software_genre law.invention Extended Kalman filter symbols.namesake Robustness (computer science) law symbols Data mining Radar business computer Doppler effect |
| Zdroj: | IFAC Proceedings Volumes. 31:489-494 |
| ISSN: | 1474-6670 |
| Popis: | The first purpose of the paper is to compare some data association methods, in order to build a multiple-target tracking system. The sensor is a Doppler effect radar, whose characteristics are known. The tracking is done by an extended Kalman filter. We present four classical algorithms : the Nearest Neighbour algorithm, The Probabilistic Data Association Filter, its extension to the multiple-target case, called the Joint Probabilistic Data Association Filter and finally the Multiple Hypothesis Filter. The comparison of their performance (robustness/autonomy/computation time) is done on simulated data, as well as on real data. Then we build a new method, which use two of the prior algorithms, in order to find the best compromise with regard to the application. This method, called Hybrid Filter, is also compared to the others. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|