An autonomous vision-based target tracking system for UAV
In this project, an autonomous vision-based tracking system is presented to track a maneuvering target for a rotorcraft unmanned aerial vehicle (UAV) with an onboard gimbal camera.
The contributions of the project are summarized as follow:
In the case of target occlusions or loss, the status of the target, i.e. loss or not, is firstly detected based on the KCF tracker, and a computationally efficient redetection method is presented. With this scheme, the UAV can track the target again when it re-appears.
An Interacting Multi-Model Extended Kalman Filtering (IMM-EKF) based target state estimator is presented to estimate states of the maneuvering target, and a nonlinear feedback control law is presented to stably track moving targets.
A computationally efficient framework implemented on onboard TK1 computer is presented for ground target tracking in unstructured environments.