Data-driven infrared small target prediction for drone swarm in complex environment

In the paper, small target prediction for drone swarm is conducted by data-driven learning. First, prediction of trajectories are generated by a LSTM network. Secondly, the angles of velocity and sight lines are calculated and normalized to generate the prediction of velocity. Finally, target prediction is generated by conducting decision fusion of the predictions of trajectory and angles. Experimental results demonstrate that the framework can fulfill the target prediction task quickly and satisfactorily.