A Multiple Model Adaptive Tracking Algorithm against Airborne Targets.
AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH SCHOOL OF ENGINEERING
Pagination or Media Count:
This thesis extends the AFIT research directed towards replacing a standard correlation tracker with a Kalman filter bankenhanced correlation tracker in a high energy laser weapon system. Airborne targets are tracked by a Bayesian multiple model adaptive filtering MMAF algorithm, which utilizes an array of infrared sensing detectors as the measurement information for two-dimensional position data. Two different target dynamics models are exercised a linear, Gauss-Markov acceleration model, and a nonlinear, constant turn-rate model. Performance analyses are accomplished via Monte Carlo simulation techniques. Extending the adaptive potential of the tracking algorithm is of primary emphasis. The effects of bending and vibration of a large space structure on the FLIRs ability to resolve target position is analyzed. Also, a performance comparisonsimulation time tradeoff is conducted with the tracking algorithm operating at both 30 Hz and 50 Hz. Sensitivity studies of adaptive responsiveness to varying target trajectories, various filter-assumed correlation times, range to pixel size relationships, and pixel size to filter driving white noise strength relationships are performed. The robustness of the multiple model algorithm is demonstrated by its ability to adapt to scenarios which it had not been previously tuned.
- *ADAPTIVE SYSTEMS
- *FORWARD LOOKING INFRARED SYSTEMS
- *INFRARED DETECTORS
- *MONTE CARLO METHOD
- PERFORMANCE TESTS
- TWO DIMENSIONAL
- AERIAL TARGETS
- Infrared Detection and Detectors
- Target Direction, Range and Position Finding
- Computer Programming and Software