The Autocopter’s Autonomous Technology

The ‘Autonomous System’ of the aircraft can be viewed as an integration of three main parts: 1) the aircraft’s stability/height, speed, attitude etc control/rotor speed control/’hard’ and ‘warning’ avoidance unit, 2) the ‘pilot’ input unit that commands the aircraft what to do either by an actual pilot (in semi-autonomous mode) or a software ‘pilot’, and 3) the flight management, route following and planning unit. As such, it is therefore a lot more than a sophisticated autopilot.

The current design is totally deterministic in what it does. In other words, a software module takes in data, uses an algorithm to compute a result and then sends that off.  The algorithm could be a polynomial neural network but it cannot, at the moment, ‘learn on the job’. That must be done as part of flight trials and be tested in the normal way for any piece of safety critical software. Horizon Helicopters will look at deep learning technology and may apply it if, and only if, the Regulator approves.

Prototyping flight control software is being provided by a subcontracted partner whose principal business is drone flight control. Horizon is planning to fly three gas turbine drones in the first year for work on main rotor and flight control proving.