This paper presents an investigation into models that can be used to train unmanned aerial vehicles (UAVs or drones) to avoid obstacles and to keep a safe distance from such. The paper describes the key components of the MATLAB/Simulink model and covers the subsystem of blocks for the UA V flight scenario. The authors have tested two scenarios based on vector field histograms (VFH) using different parameters, whereby the accuracy of tracing flight waypoints varied. Simulation results are shown in graphs. The paper further shows how the ROS simulator can use learning algorithms that could be tested on real aircraft.