What makes a great fixed wing drone? If you've done even a small amount of research you'll have realised that they come in all shapes and sizes. Some of them can almost be casually tossed into the air before flying off on a pre-planned mission, others need a nice, long, smooth runway to get airborne. They are better suited to some roles than multi-rotors are. For example they can travel faster and for longer than a quadcopter can.
Types of wings and airframes
The first and most important consideration for a fixed wing drone is the wing type and the airframe configuration. Just because it's fixed wing it doesn't mean you are limited to a traditional layout with a prop and motor at the front, two straight and parallel wings and a traditional tailplane and fin.
The requirements of a drone for say mapping or survey work are different to the needs of a small passenger plane. The payload, the runway (if you're lucky enough to have one), the air speed, stall speed, endurance, size, portability and foldability are all factors that designers need to consider.
Very often developers of this type of craft decide to keep things really simple, using variations on a delta or a swept wing design with no separate tailplane and just a couple of winglets. They're often made from lightweight polystyrene with a few reinforcements here and there. Some are tougher, with carbon fibre frames. They tend to have one pusher prop and don't bother about an undercarriage, relying instead on belly or parachute landings. The theory is that there's less to get damaged in rough take-offs and particularly landings. They can also be cheaper to produce.
They usually have a central payload bay/fuselage with detachable wings for ease of transport and not all of them are small. Some have wingspans measured in metres. However the flight characteristics of flying wings are not usually quite as good as a more conventional design even though they're a practical proposition for rugged work.
We've been testing our equipment on a simple swept wing design, using an off the shelf Skywalker X-8 with help from Ben Wilkinson of Aerial Exposure. The SkyWalker is a simple foam structure with removable wings given rigidity by carbon fibre tubes and held in place by plastic clips. It has a removable payload hatch on the back, giving easy access to the batteries and avionics as well as the camera gear. We've been using it to test out different components.
Although the wingspan is more than two metres (212 cm) it's still easy to hand launch. Even with a prop at the back you can hand launch safely by programming the motor to start up at a certain speed or a fixed time after launch. With this sort of configuration you're probably looking at a flight time of between 40 and 60 minutes. Bigger wings and batteries and more powerful motors boost the payload capability.
The SkyWalker X8 test aircraft
Requirements for a good fixed wing drone
So what are the requirements for a fixed wing UAV design? You need it to have:
- A wide range of operating speeds
- A slow stall speed
- Gentle stall behaviour
- Long endurance
- A large and stable payload capacity
- Unobstructed field of view for sensors
- Multiple launch modes (hand, catapult and smooth/rough strips)
- Belly landing capability including payload protection
- Easy assembly and break down
- Easy to transport
That's quite a tall order but our research is leading us towards an unusual and more complex wing shape, tailplane and motor configuration for this type of aircraft which we are confident will fit the bill. We're talking a payload capacity of up to 4kg. Watch this space for more information soon.
Autopilots for fixed wing UAVs
For an autonomous aircraft you're going to need an autopilot. 3DR Pixhawk is one of the better known and possibly the cheapest option. It's an opensource, versatile piece of kit that can be used for fixed wing, multi-rotors, helicopters, cars, boats and any other robotic platform that's capable of moving. It's also the one that seems to be working well in our test aircraft.
The Pixhawk autopilot
ArduPilot Mega is another open source autopilot that's based on the Arduino Mega platform. The software is constantly updated by a team of around 30 core developers, supported by a community of more than 10,000 members. For many people the opensource technology of Pixhawk and ArduPilot Mega is what makes them so attractive.
These two autopilots are quite often linked to a Raspberry Pi mini computer, usually on the ground but sometimes on the aircraft itself to help share the data load. Only this week version 3 of the Pi was announced and it has more processing power.
Micropilot is one of the manufacturers of high end autopilots that are usually found in ready to fly UAV packages. The flight controller is usually the most expensive item in this type of set-up however MicroPilot do provide their own Horizon Ground Control Station software as part of the deal.
Big aerospace, defence and technology manufacturer Lockheed Martin also produces a UAV autopilot - the Kestrel. It links up with Virtual Cockpit, a 3D graphical user interface with waypoints and streaming maps.
Onboard sensors and other relevant equipment
For maximum flexibility a fixed wing UAV needs to be able to carry a wide range of sensors. If it can carry more than one at a time that's even better. Combinations of standard optical, infra red, red edge, hyperspectral and LiDAR can all be loaded depending on the job. Imagery can be used for crops surveys, 3D modelling, volumetrics and thermal
3 images of the same field - normal, near infrared and combined. Credit: Agribotix
Plenty of room to stow them and plenty of scope to move them around to alter the centre of gravity are both important. The cameras usually point straight down so vulnerable lenses need to be lifted out of harm's way by an undercarriage or shielded by a protective flap that can be triggered. A good 2 axis stabilising gimbal for sensors is a vital component, although some fixed wing UAVs get round the problem by stopping the aircraft's motor momentarily while the picture is taken. The Sensefly eBee uses this method.
Of course the aircraft doesn't just gather image data. It has to link up with the GPS so that each image is geotagged. Getting high quality images is only a small part of the equation. If there is inaccurate geotagging, incorrect overlap of the images or ground control points weren't used then the precision of the data will be flawed.
Ground Control Points are markers with accurately recorded GPS coordinates that will be visible from the air. This helps to align the captured images with the site that's being surveyed. Usually the client's surveyor will provide the GCPs.
Good flight planning software is a vital part of the operation too. If you feed in the type of camera, sensor and lens it will calculate the optimum height, speed and image overlap for your flight and then plot the best route to achieve it. What you are aiming for is high quality resolution of around 1cm per pixel. Once that's set up all you have to do is launch it and let it fly its mission.
The Pix4D Capture app for pre-planning your route
Software for image processing
The other half of the software equation is data processing. This has to collate the masses of images and location data and then interpret them and convert them to the format required by the customer.
In its simplest form it could mean the stitching together of a set of photos into a large 2D image. Even then the software has to allow for overlaps and slight changes in image orientation. Then there's 3D mapping and building surveying as well as volumetrics - the ability to measure for example mounds of earth or the amount of material that's been extracted from a quarry or opencast mine.
An orthomosaic map - Conservationdrones.org
Software can also be used to interpret information about crops or forestry plantations. Apart from assessing plant health and the level of watering it can also measure crop height and the number of plants per row.
Some of the main players in this market are Pix4D, Agisoft, Datamapper and Drone Deploy. Some supply compatible software for mission planning and capture while other deal just with image processing. Increasingly the cloud is used to upload captured data for the operator or the client to process back at base.
For all of this sort of work a solid, reliable and stable aircraft is needed. It must be easy to transport and assemble onsite and be capable of short takeoffs and landings in a wide range of conditions. We think our research and development work is leading us to the ideal aircraft for this type of role. We can't wait to share our design with you in the not too distant future.