Cascaded PID Control System for UAV with Gain Factor Prediction Using ML

Drones are not inherently stable, necessitating the use of a flight controller. If the UAV is properly tuned, the drone will fly steadily; otherwise, it won't. Hence, we have used a PID (proportional, integral, differential) controller for a stable flight. A well-functioning PID controller should enable amazing climbs and long-range flights. But, when used singly, PID controllers can provide poor performance, resulting in a long settling time, overshoot, and oscillation. Here, we propose a new approach to maneuver UAVs using a PID control system and overcome the shortcomings of using PID controllers in UAVs. This disadvantage is resolved using the Machine Learning polynomial regression model. The gain factors in a PID control system, which is otherwise ideally constant, should be changed in order to reduce the minor instabilities for a smooth flight. Our method has been elaborated and illustrated with suitable diagrams in the following work. When simulated in Gazebo on a Robot Operating System (ROS), our technique is proven to be successful.