Estimation and Guidance Strategies for Vision-based Target Tracking

This paper discusses estimation and guidancestrategies for vision-based target tracking. Specificapplications include formation control of multiple unmannedaerial vehicles (UAVs) and air-to-air refueling. We assumethat no information is communicated between the aircraft,and only passive 2-D vision information is available tomaintain formation. To improve the robustness of theestimation process with respect to unknown target aircraftacceleration, the nonlinear estimator (EKF) is augmented withan adaptive neural network (NN). The guidance strategyinvolves augmenting the inverting solution of nonlinear line-of-sight (LOS) range kinematics with the output of anadaptive NN to compensate for target aircraft LOS velocity.Simulation results are presented that illustrate the variousapproaches.