On the challenges facing airspace integration of small UAVs: autonomous navigation and control in a turbulent urban environment

Future airspace integration of autonomous unmanned aerial vehicles will require a comprehensive understanding of the myriad of environmental conditions that could impact vehicle performance. Regulatory bodies, such as the Civil Aviation Authority and Federal Aviation Administration, have yet to be satisfied with the autonomous flight capabilities of the types of aircraft that might one day occupy an urban airspace. As well as being a novel guidance, navigation and control problem, it is also a consideration of external environmental effects, such as structure induced turbulence, and its impact on rotorcraft stability. The importance of accurate flowfield modelling is established, with both temporal and spatial characteristics identified as being critical to a realistic assessment of vehicle response during flight simulations. Adverse aerodynamic effects are transmitted to the rotorcraft platform, via its propellers, by utilising a widely adopted combination of momentum and blade element theory. This approach is further improved by the derivation of polynomial lift-curve slope expressions. Recognising the computationally prohibitive nature of large eddy simulation, when applied to urban flowfield generation, unsteady aerodynamics are approximated by extracting turbulent kinetic energy values obtained from Reynolds-averaged Navier-Stokes simulations, and incorporating them within a Dryden statistical turbulence model. This approach is used to define a critical turbulent kinetic energy value for a quadrotor, allowing the urban environment to be characterised in terms of turbulent hot spots. Treating these areas as impassable within 2D and 3D A* algorithms, safe paths through a time-averaged flowfield are generated, with lower altitude flight preferred for operations where the impact of turbulence can be minimised. Illustrating the complexity of the airspace integration problem, studies of realistic urban environments demonstrate a relationship between prevailing wind direction and the spatial distribution of turbulent kinetic energy. Model predictive control was employed to demonstrate the effectiveness of a receding horizon approach, when mitigating the effects of discrete gusts and statistical turbulence on a rotorcraft platform. Without gust previewing, the quadrotor experiences saturation of the actuator input rates, as it attempts to reject the impact of a discrete gust. When gust previewing is active, the vehicle responds in advance to progressive gust immersion by generating a large angular velocity which is arrested by the oncoming gust. This results in reduced displacement of the vehicle, associated with a smaller input range and near elimination of rate saturation. Active gust previewing was found to be less effective when applied to statistical turbulence, however, the model predictive control algorithm was still capable of rejecting large variations in the mean flow