PROJECT TITLE :

Reinforcement Learning-based Collision Avoidance and Optimal Trajectory Planning in UAV Communication Networks

ABSTRACT:

In this paper, we investigate optimal trajectory planning for unmanned aerial vehicle (UAV) communication networks and propose a reinforcement learning approach to collision avoidance as a means of avoiding collisions. Specifically, it is the responsibility of each UAV to deliver objects along the forward path and collect data from a variety of different Internet of Things devices on the ground along the backward path. We use reinforcement learning to assist unmanned aerial vehicles (UAVs) in learning how to avoid collisions without having prior knowledge of the trajectories of other UAVs. In addition, we use optimization theory to determine, for each UAV, the shortest backward path that ensures data collection from all associated IoT devices. This path is found by retracing our steps forward. A no-return traveling salesman problem is formulated and solved by us in order to obtain an optimal visiting order for Internet of Things devices. We formulate and solve a series of convex optimization problems in order to obtain line segments of an optimal backward path for heterogeneous ground IoT devices, and we do this by starting with a visiting order as our starting point. For the purpose of justifying the utilization of the suggested method, we make use of both analytical results and simulation results. The results of the simulation show that the approach that was proposed is superior to a number of other approaches that were considered.