Treffer: Developments of intelligence-based quadcopter drone for object detection application

Recently Quad-copter have been an open research area because of their simplicity construction, low in cost, the ability of manoeuvrability including hovering, vertical take-off and landing. In addition to that quad-copter can accomplish many tasks such as surveillance, search and rescue, delivery and military application. However quad-copters are inherently nonlinear and under actuated at the normal condition this results increase the difficulty of the researcher to stabilize and to control both the translational and rotational dynamics. In this research project the mathematical modelling of the quadcopter with the account of gyroscopic effect and aerodynamics effects are considered and formulated using Euler Lagrange formula in addition. In additions we develop two controllers the first one is PID controller used to stabilize the quadcopter around the desired positions as well as a desired Euler angles because controlling 6 degrees of freedom with only 4 control inputs is impossible the second one is the reinforcement learning controller (RLC) in the RL controller the quadcopter is trained to achieve a desired altitude . The RL controller is implemented in python 3.7.9 and the PID controller is implement MATLAB-SIMULINK. For hardware implementation we used Electronic Speed Controllers (ESCs) which command by NVidia Jetson Nano AI platform to govern the angular velocities of each quadcopter motor. Finally, we deployed an object detection algorithm from the live camera for the feasibility of our tasks.