The original focus of this simulator was on accurately simulating rotary-wing UAV dynamics with added lumped parameter representations of the aerodynamics for course design and exploratory research. These aerodynamic effects, listed below, are negligible at hover in still air; however, as relative airspeed increases (e.g. for aggressive maneuvers or in the presence of high winds), they quickly become noticeable and force the student/researcher to reconcile with them.
As RotorPy continues to grow, the focus is now on building a realistic dynamics simulator that can scale to quickly generate thousands (or even millions) of simulated rotary-wing UAVs for applications in deep learning, reinforcement learning, and Monte Carlo studies on existing (or new!) algorithms in estimation, planning, and control.
The engine is designed from the bottom up to be lightweight, easy to install with very limited dependencies or requirements, and interpretable to anyone with basic working knowledge of Python. The simulator is intended to gain intuition about UAV dynamics/aerodynamics and learn how to develop control and/or estimation algorithms for rotary wing vehicles subject to aerodynamic wrenches.
The following aerodynamic effects of interest are within the scope of this model:
Parasitic Drag – Drag associated with non-lifting surfaces like the frame. This drag is quadratic in airspeed.
Rotor Drag – This is an apparent drag force that is a result of the increased drag produced by the advancing blade of a rotor. Rotor drag is linear in airspeed.
Blade Flapping – An effect of dissymmetry of lift, blade flapping is the motion of the blade up or down that results in a pitching moment. The pitching moment is linear in the airspeed.
Induced Drag – Another effect of dissymmetry of lift, more apparent in semi-rigid or rigid blades, where an increase of lift on the advancing blade causes an increased induced downwash, which in turn tilts the lift vector aft resulting in more drag. Induced drag is linear in the airspeed.
Translational Lift – In forward motion, the induced velocity at the rotor plane decreases, causing an increase in lift generation. Note: currently this effect is NOT modeled in the thrust produced by the rotor.
Translational Drag – A consequence of translational lift, and similar to Induced Drag, the increased lift produced in forward flight will produce an increase in induced drag on the rotor.
Ultimately the effects boil down to forces acting anti-parallel to the relative airspeed and a combination of pitching moments acting parallel and perpendicular to the relative airspeed. The rotor aerodynamic effects (rotor drag, blade flapping, induced drag, and translational drag) can be lumped into a single drag force acting at each rotor hub, whereas parasitic drag can be lumped into a single force and moment vector acting at the center of mass.
What’s currently ignored: any lift produced by the frame or any torques produced by an imbalance of drag forces on the frame. We also currently neglect variations in the wind along the length of the UAV, implicitly assuming that the characteristic length scales of the wind fields are larger than UAV’s maximum dimensions.
RotorPy also includes first-order motor dynamics to simulate lag, as well as support for spatio-temporal wind flow fields for the UAV to interact with.