Reinforcement Learning for Autonomous Drones are quite appealing topic. It includes training a model. However, these approaches tend to be sensitive to mistake by the teacher and don’t scale well to different environments or vehicles. To the present finish, we have a tendency to propose a standard network architecture that decouples perception from management, and is trained mistreatment empiric Imitation Learning, a novel imitation learning variant that supports online coaching and automatic choice of the best behavior from perceptive multiple academics. We have a tendency to apply our planned methodology to the difficult downside of remote-controlled aerial vehicle (UAV) sport. We are developing a machine that permits the generation of enormous amounts of artificial coaching information (both UAV captured pictures and its controls) and conjointly permits for online learning and analysis. Our proposed system will intelligently navigate in indoor and outdoor environment. It also involves association of assistants like Alexa, Google Assistant, Bixby etc. The Integration of Mapping with the help of SLAM algorithms through IMU sensors. The System relies on End device and Cloud For Computation. User Control is provided from Android Device. User Tracking and Waypoint navigation with the autonomous flight is additional enhancements to the Drone.

Keywords : Artificial Intelligence, Machine Learning, Unmanned Aerial Vehicle, Reinforcement Learning.