Despite the current advancements in the navigation technology of drones, ensuring high-precision autonomous navigation, especially in outdoor and unknown environment, is still a huge challenge for real missions covering long range. In the outdoor environment, drones are subjected to inaccurate position and velocity errors due to the lack of precise positioning systems. In addition, they also encounter large wind disturbances, making the task of precision navigation significantly difficult. In addition to these, for collision avoidance and successful recovery (landing), situational awareness is a must and that needs to be done in real-time to be useful for the vehicle navigation. It can be addressed in a limited sense using Artificial Intelligence (AI) techniques. Nevertheless, many existing AI algorithms prohibit real-time implementation. In addition to the situational awareness, powerful nonlinear and optimal guidance laws catering to large perturbations are also needed, which can demand small control efforts. The objective of the research is to achieve precise autonomous navigation in drones in complex setting like navigation in outdoor and unknown environments, takeoff and landing on a moving platform. Developed concepts will also be experimentally demonstrated from field trials.

Software in the loop simulation setup for a Quadrotor drone using PX4-ROS (Robot Operating System) environment. Since, achieving precise landing needs precise estimation of landing position relative to the drone body frame. Estimation of target position using multiple April tags to increase the ran ge of detectability of an April-Tag. This data is used to fuse with GNSS units to precisely estimate the position of drone relative to target frame

Study of Guidance in windy conditions:

Thus, minimum jerk based guidance strategy gives smooth trajectories and the boundary conditions on the position, velocity, and acceleration enables soft landing on the target. In order properly validate the algorithm in realistic conditions especially in outdoor conditions, wind is given as a disturbance into the system. Due to this the formulation of the guidance is modified by including the aerodynamic acceleration before passing to the inner control loop. Wind speed is modelled by selecting different harmonics of a Fourier series. Assuming the wind data is known, the guidance strategy is simulated for various wind conditions and studied. 

This correction for aerodynamic acceleration depends on the wind speeds which are typically unknown or cannot be precisely estimated. Since, estimation of wind requires typical airspeed sensors which are inaccurate in low-speed vehicles like a quadrotor, other methods must be employed to obtain the wind information.