Random disturbance presents a reliability and a safety defy for quadrotor control, This research demonstrates an adaptive linear quadratic Gaussian (LQG) control of quadrotor, exploiting a novel faster full state observer based on an extended Kalman filter enhanced by the Madgwick method, using data fusion of multiple asynchronous sensors, subjected to track a remotely generated Spline trajectory for obstacle avoidance. The dynamics model of the quadrotor was derived using Newton Euler formalism; furthermore, its linearization was processed by the Jacobian matrix at every estimated state. The enhanced state observer is essentially based on a continuous-discrete nonlinear Kalman filter combined with the optimization of Madgwick method for quaternion orientation. The approach relies on flight dynamics predictions and gets updated by the onboard measurement of sensors at different feeding rates. A discrete linear quadratic tracker was developed for best tracking performance and robustness while avoiding collision with predefined static obstacles. Quaternion orientations and attitude and heading reference system were validated by experimental tests at less than one degree of precision error, whereas the proposed adaptive LQG control of quadrotor was simulated for tracking control and validated in Simulink environment in the presence of Gaussian random perturbations, LQG performance has been compared to the linear inner-outer looping PID, integral backstepping (IBS), and decentralized fuzzy logic control (FLC) strategies. Obtained results validated the effectiveness and robustness of the opted methodology.