Trajectory tracking control of quadcopters under tunnel effects

There are many potential applications to utilise aerial robots in hazardous tunnel-like environments. For example, aiding human operators with inspections of small railway culverts or mineral mappings of mining tunnels. Nevertheless, such confined environments pose many challenges for quadcopters to navigate through. Suspended dust particles, poor lighting conditions and featureless/excessive features in the surroundings make localisation difficult. Furthermore, the fluid interactions between the rotors’ downwash and the surfaces of the surroundings create aerodynamic disturbances, which threaten the quadcopter's stability and increase its risk of collision in the restricted confined space, not to mention the longitudinal wind gusts. This paper presents our findings on the characteristics of these aerodynamic disturbances, the Tunnel Effects for quadcopters, in a 1.5m(W) x 1.5m(H) square cross section tunnel through a series of experiments. A semi-autonomous system is proposed with self-stabilisation in the vertical and lateral axes while a pilot provides commands in heading and the longitudinal direction of the tunnel for performing required tasks such as tunnel wall inspections. We propose a cross-sectional localisation scheme using Hough Scan Matching with a simple kinematic Kalman filter for providing reliable vertical and lateral position information. An integral backstepping (IBS) controller is designed and implemented to enable quadcopters to robustly fly in tunnel-like confined environments. The proposed system is tested in simulated tunnel environments and a real railway tunnel with various reference trajectories, and the IBS controller has shown superior tracking performance in comparison with a PID controller despite of the existence of the Tunnel Effects.