Geintra

There are many different solutions for tracking multiple objects and many of these solutions involve probabilistic algorithms, which have been fully tested as the best solution in tracking tasks. In this thesis a multiple tracking algorithm based on the Kalman Filter and the Probabilistic Data Association Filter is developed. The algorithm is part of an obstacle avoidance system in an autonomous robot. The measurements used as input to the tracking algorithm come from a stereo-vision system that detects objects in the robot’s environment. The robustness and adaptability of the tracking algorithm is increased by the use of a validation/removal algorithm. The algorithm is capable of initiating tracks, accounting for false reports, and removing tracks, accounting for missing reports.