What is an unmanned aerial vehicle?
Recent years have seen an increase in the deployment of aircraft that do not have a manned pilot on board. These unmanned aerial vehicles (UAV) are also commonly known as drones, unmanned aircraft systems (UAS), or remotely-piloted aircraft systems (RPAS). For the purposes of the following descriptions, these terms will be used interchangeably. Drones of all shapes and sizes have been used in the military for many years. Advantages brought by increasingly efficient and adaptive manufacturing procedures (such as 3D printing), and the miniaturization of electronic components; have since made it much more viable to deploy UAVs in civilian contexts. Access to the technology is now such that it is straight-forward for the general consumer to purchase a UAV, practice flying it and mount cameras and other sensors to take pictures of the environment around them.
UAV navigation requirements
A UAV can be a fixed-wing, single or multi -rotor aerial platform that is remotely-controlled by a ground-based human operator. Currently, for civilian use in many of world’s flight jurisdictions, the UAV will need to be flown within the unaided eyesight (visual line of site – VLOS) of the operator. Such regulations are still evolving and there are now an increasing number of situations where beyond line of site (BLOS) operations are now being permitted. When considering professional and high-end consumer grade UAVs, most UAV platforms will make use of a global positioning system (GPS). Many will also include at least a low-grade inertial measurement unit (IMU ) within its flight controller for flight navigation and control.
GPS is often the generic term used to describe global navigation satellite system (GNSS) technologies, which are used to calculate a position on the earth’s surface by way of timing information received from a network of GNSS satellites that continually orbit the earth. The satellite constellations accessed by the GNSS systems might include the original US-led Global Positioning System (GPS) project, the Russian GLONASS constellation, or one of the other constellations under development such as the European -led GALILEO, or China -based BeiDou projects. With access to GNSS technology is the ability to automatically compute and locate the position of the UAV in line with a real-world coordinate system.
An IMU is comprised of an assembly of gyroscopes and accelerometers. On a UAV platform these IMU components are increasingly manufactured using micro-electromechanical (MEMS) technology . It is the IMU that will provide data related to the linear acceleration of the UAV on three axes, together with measurements pertaining to the rotation of the UAV in terms of roll, pitch and yaw.
The flight controller on the UAV will use data from the GPS to provide coordinates of where the UAV is at a particular point in time. Data from the IMU, will tell the flight controller whether the UAV is level, if it is rotating, and essentially how stable it is during flight. If the a flight controller uses both inertial and GPS information, necessary feedback to both the UAV and its operator can be provided for safe operations in both VLOS and BLOS operating scenarios.
Challenges of aerial mapping
Aircraft-mounted cameras have been used for surveillance or mapping purposes since the mid-1900s. When images from aerial cameras are used for the purposes of using photogrammetry to map topography or land cover, or calculating slope angles or mine volumetrics; it is necessary to not only understand the specifications (internal model) of the camera or sensor used, but also:
- How the exterior orientation of the imaging sensor changes (i.e. how the sensor moves relative to the ground).
- The effect of distortions that inherently three-dimensional ground geometry has on the measurements and distances in mapping output when imaged from above.
The aircraft’s navigation systems provide information on the routing and part of the World that the aircraft is flying over. For the creation and update of mapping, the location and footprint of the camera’s photographs have needed to be anchored via known coordinates of features in the landscape (ground control points – GCPs) that are seen in the photographs. Undulations in the topography of the ground can be accounted for by applying aerotriangulation equations to both tie photographs together and associate features in the image with the x,y,z coordinates of the GCPs. However this is both a manually intensive and costly process, where the quality of result will be dependent upon factors such as the number of GCPs used and the patterns that the GCPs are laid out in.
It is this method of georeferencing to the coordinates of known GCPs that has mainly been used to provide survey control to UAV aerial photography. As UAV platforms have been increasingly aimed at ever wider markets, computation systems and therefore costs have been reduced to provide more favorable pricing points for the platforms themselves. Methods to further refine GPS readings on the UAV platform, such as receiving atmospheric and timing corrections from a local base station via either a real-time radio link (real-time kinematic GPS) or via a post-processed differential GPS workflow, are now commonplace. Nevertheless, this still does not negate the use of GCPs, and it can be easy to overlook the time required to establish survey control or undertake these required aerotriangulation procedures.
Direct georeferencing
Many of the distortions seen in aerial photography can be attributed to changes in exterior orientation caused by the fact that airborne platforms rarely move in a completely stable and level manner. The platform itself will bank around corners, atmospheric conditions will cause it to pitch and yaw. Even if the camera is fitted to a self-leveling gimbal, in extreme circumstances a camera may not always be oriented perpendicular to the ground surface. Additionally, the earth is not flat. Undulations on the ground below the aircraft (hills, mountains, valleys, buildings), together with variations in altitude of the aerial platform will cause changes in perspective in the image captured, and as a consequence cause distortions in any distances that are measured.
To derive reliable mapping products from aerial photography, maps will be digitized from overlapping stereo photographic pairs, or processed orthophotographs. The aim of both of these techniques is to essentially normalize the effects of undulations in the landscape, to produce a mapping product where distances are true. To undertake mapping in a modern automated manner it is not only necessary to collect data on the coordinates of where the photograph is being taken, but also to understand the camera model, together with data related to the overall trajectory of camera movement and its orientation parameters at the specific point in time when the photograph is taken.
Orientation information is derived from an IMU, and more specifically from a computational module that statistically processes both location data from the GPS system (onboard and from localized base-station), together with orientation information from the IMU to make an overall best estimate assessment of the camera’s trajectory. It is this computational module that enables dependable mapping products to be produced while the aerial platform is moving, and it is these systems that are referred to as inertial navigation systems (INS). Mapping-grade INS systems have been utilized widely on manned airborne platforms since the late 1990s . On many of the airborne mapping sensors deployed on manned platforms the IMU component of the INS is now integrated onto the sensor itself.
Direct georeferencing not only provides the capability to efficiently collect aerial photography, but also apply to all other types of mapping sensor such as light detection and ranging (LiDAR) systems and other imaging sensors (for example, those that measure in hyperspectral bands).
Aerial photography and other imaging sensors
UAVs are typically flown with lower cost cameras than for manned photography operations, often these cameras were not originally designed for airborne use. Accordingly, it is rare for these cameras to already incorporate sophisticated INS systems for deployment on the UAV. This means that it is even more important to consider incorporating an INS onto a UAV platform to improve the efficiency and quality of a mapping project. Ideally the INS will be mounted as close as possible to the mapping sensor in order to get the best estimate of its position and orientation of the mapping sensor throughout the mission. On manned airborne platforms, due to the size of the aircraft the INS system might still be situated some distance away from the camera mount. A relative distance, angle and orientation then needs to be calculated between the INS body and the camera frame, providing what is known as the lever-arm vector.
Due to their reduced overall size, the distances between the INS body and center of the camera frame on a UAV are likely to be much smaller than on a manned platform. Often the mapping sensor will be mounted directly on, or below the INS; nevertheless calculating the lever-arm vector will further improve the direct georeferenced results.
Fixed-mount or gimbal?
If the camera is on a fixed-mount, the movement of the UAV body as measured by the INS will directly translate to movement in the spatial position of the camera. If the camera is not mounted in a conventional nadir (i.e. focused directly below the UAV), but at an oblique angle, then proper calculation of the lever-arm vector to the INS will account for the angle that the camera mount is on and enable direct georeferencing of the image.
Frequently gimbal-style mounts are used on UAVs to stabilize the mounting angle of the camera to the ground, independently of movement in the UAV platform. In theory, this means that while the UAV may roll, pitch or yaw, the camera should remain level to the ground. What the gimbal does not do is account for dynamic changes in altitude or those same motions on the UAV platform; that will also have an impact on the position of the camera in space. Nor does a gimbal negate the requirement for an INS to calculate an overall best estimate of position throughout the survey. Accordingly, even if the camera is mounted on a gimbal, utilizing an INS still provides the best direct georeferenced result on the mapping project.
Lidar sensors
Lidar systems bring many advantages for some types of mapping projects, such as those concerned with forestry and mining. Unlike camera sensors that typically capture data passively across image frames, lidar systems are active systems. A lidar collects data by emitting individual pulses of light and calculating the time that it takes for that pulse of light to return to the lidar sensor. Many thousands of pulses of light per second will be emitted across a swath, known as a scan line. Since the speed of light is known, the distance of each pulse to its target (for example, the ground) can be calculated. To calculate an x,y,z coordinate position of the target also requires that the exact position and orientation of the aircraft to be known and for it to be dynamically assigned to each these pulse returns. As a result direct georeferencing of UAV mounted lidar is essential, using a high-grade INS system. As with other mapping sensors such as aerial cameras, best results are obtained when the INS and lidar system are mounted together on the UAV and for lever-arms to be calculated that account for the mounting angle of the lidar.
Why inertial navigation systems are essential for UAV-based mapping operations
While they are smaller in size, can be implemented at a lower initial capital cost, and are typically operated over smaller areas than manned systems; UAV based mapping operations still need to operate efficiently to be able to better capitalize on the advantages of a lower cost platform. Typically, a UAV based mapping project will be delivered at a lower price-point than a manned operation. To stay competitive, UAV operators need to be sensitive to the costs of time spent onsite to conduct survey work. Traditional georeferencing tasks utilizing GCPs are well known for being manually intensive. Regardless of the sensor used, similarly to experiences learned in manned airborne survey operations, direct georeferencing using an INS provides a means of accurately collecting mapping data from a UAV, and is essential for utilizing advanced sensors such as LiDAR.
