UAV – Virtual Migration Based on Obstacle Avoidance Model

In recent years, the obstacles avoidance technology of unmanned aerial vehicles has been developed rapidly. It takes a lot of manpower to control un-manned aerial vehicles, so many researches use reinforcement learning to make unmanned aerial vehicles fly autonomously. In the real environment using rein-for cement learning to train aircraft is an expensive and time-consuming work, because reinforcement learning is a way to learn from mistakes, so there are often bumps in the learning process. In Wu’s research, they trained a good model, but the realistic environment and simulation environment differs very big, so we will train this model again and transferred to the real environment, makes unmanned aerial vehicle in the realistic environment can use cheaper and quickly achieve the same task.