The control of equipment such as camera gimbal, Vertical Take-Off and Landing (VTOL) and Load Transporting System (LTS) on Unmanned Aerial Vehicle (UAV) with its own flight control directly affects the performance of the mission in tasks such as tracking the target along the specified path and leaving payloads on the targets specified in the dangerous areas. In this study, neural network based real-time control of a hexarotor UAV is performed so that the payloads on the targets determined by path tracking can be left with minimum error. The Nonlinear AutoRegressive eXogenous (NARX) model of the UAV is obtained after the flight data are passed through the preprocessing, feature extraction and feature selection stages. The obtained neural network model is embedded in the flight control card to realize real time path tracking of the UAV. The three payloads in the cubic structure are both transported by the originally designed LTS and left with the help of LTS to targets on the path. Environmental testing is conducted taking into account the limitations of the physical properties of the LTS and specified path tracking on the autonomously moving UAV, and the impact on proposed NARX control algorithm's mission performance is examined.