Unmanned Systems Technology 004 | Delair-Tech DT18 | Autopilots | Rotron RT600 | Unmanned surface vehicles | AMRC | Motion control | Batteries

33 the importance of demonstrating that the autopilot operates safely. In the past, such issues have meant developing custom, proprietary autopilot systems for specific systems and using particular software and hardware architectures. With the increasing popularity of unmanned aerial systems in areas such as agricultural monitoring, there is however a move to standardise the autopilot function and separate out the software for a given application. Functions of an autopilot Quite what the control system and autopilot are is also becoming more complicated. For example, the autopilot for a large UAS acting as an internet base station in the sky will have to keep the craft in position above the ground station with high accuracy to ensure that the laser communications downlink to the ground is maintained. It will have to cope with changing wind speeds and air density to do so, and the autopilots of neighbouring craft will have to monitor the position of each neighbour to maintain the comms links between them. All of this requires constant and accurate control of the electric motors and control surfaces of each craft. So the autopilot has to handle the automatic control of the craft, from the operating mechanisms, engine and stabilisation, to the links with a ground station. It also has to ensure that any failures are handled effectively, for example protecting a £1 million camera from a serious crash. Whether it also has to handle the payload control is a key question for system designers. Some developers provide an application programming interface to the autopilot to allow additional functions to be added by an operator or external supplier. For example, this could be linking the control of a camera on a gimbal to maintain a constant view regardless of the heading of the aircraft, which could well require a close link to the autopilot. Similarly, the data link back to a ground station could also be used for transmitting data from the payload, so an interface to the autopilot would be needed. However, the key requirement for an autopilot is safety, whether it’s used in a small quadcopter or a large military UAV. The safety requirements, and therefore the specification of the hardware and software, are determined by the regulatory environment. The requirements for an airliner carrying hundreds of people are quite clearly defined; by contrast, those for an unmanned system that could hit the airliner and cause a crash are, as yet, undefined. These requirements are reflected in the design of the autopilot system in several ways, the first of which is the speed of response, which can vary according to the speed of the UAS. For example, a quadcopter moving at a Autopilots | Focus There is now a move away from custom autopilot systems towards standardising the autopilot function and separating the software for a given application Unmanned Systems Technology | Autumn 2015 New ways to separate the autopilot and the processor for the application software (in red) allow greater standardisation of hardware and software (Courtesy of Airware) The Kestrel autopilot uses Dynamic C to encode its control algorithms (Courtesy of Lockheed Martin)