Understanding flight dynamics
When undertaking any form of flight, it is important to understand the orientation of the aerial platform, the nature of its motion and the direction in which it is heading. In the context of unmanned aerial vehicles (UAV), these characteristics of flight might be understood by the pilot simply observing the UAV platform’s motion when flying within visual line of sight. To reduce human error when flying under manual operation, a range of sensors on the UAV platform may provide feedback to the pilot’s flight controller. In other methods of flight, information related to flight motion, alongside location information from global navigation sensing systems such as global navigation satellite system (GNSS or GPS) becomes a necessity during automated autopilot operations.
The type of system that is typically used to measure this motion, is an inertial measurement unit (IMU). Typically composed of gyroscopes and accelerometers, information from the IMU characterises the dynamics of a platform’s motion in terms of a continuous stream of data related to the linear acceleration of the vehicle on three principal axes, together with rotation parameters along three principal axes (pitch, role and heading). Many UAV flight control systems will include a form of IMU to provide the necessary feedback to both an autopilot system and the pilot’s flight control software on the ground. When using a UAV to collect data for survey purposes though, a data stream from the IMU only provides part of the story.
Flight dynamics and collection of aerial data
When it comes to survey, it is important to understand the relationship between the UAV platform, the external geometry of the imaging sensor mounted upon it, and the ground below. ‘The world is not flat’, even though the output of a survey might be a two-dimensional plan; landscapes and the surveys of them are inherently 3D. At its simplest level, it needs to be known whether undulations seen in the ground below are a function of the landscape, or undulations in the UAV’s flight path. To collect data that is to be used as part of a survey, a UAV will be fitted with an imaging sensor that might be a camera or laser scanning (LiDAR) system.
In the case of aerial photography, the effect of ground height needs to be calculated to best project what is seen on the ground onto what is essentially a ‘flat photograph’ in the case of an orthophotograph.
When accuracy tolerances are frequently measured in a couple of meters or (more usually) centimeters, it can be important to calculate the even smallest of angular deviation in the movement and direction of the imaging sensor that is mounted on the UAV. In the context of laser scanning, data is collected by a pulsing laser sensor that measures how long it takes to receive a return after that pulse has hit a surface on the ground. This return might either be the top of a surface structure, or the ground itself, and returns in the direction of the sensor. Laser returns are collected over an area by a function of the fact that the laser sensor is often spinning (essentially ‘spraying’ laser pulses along a plane), and the UAV platform that the sensor is mounted to is also moving.
In a previous technical note ”Why is an Inertial Navigation System (INS) important for unmanned aerial vehicle (UAV) survey and mapping applications?”, differences between passive (via ground-control points (GCPs) only) and direct georeferencing techniques have been described. Direct georeferencing utilizes an inertial navigation system (INS) to provide a more efficient and more consistent workflow for assigning spatial coordinates to the data collected by the UAV. Due to cost-considerations and current state-of-the-art of lightweight cameras; even though phased-out in manned aerial survey 10-20 years ago, passive georeferencing is still common-place in UAV photogrammetry operations due to the consumer-grade cameras that are still used. However, if in a manner that is similar to which aerial photogrammetry from manned aircraft today that direct georeferencing is employed, it would enable UAV photogrammetric surveys to be undertaken with need for fewer GCPs. In the case of LiDAR, utilizing an INS is a necessity as coordinates of typically 100,000 points per second need to be calculated individually.
An INS includes similar gyroscopes and accelerometers that can be found on an IMU to calculate moving dynamics of the UAV, together with a computational system that can take as input GNSS data, and applies a series of statistical filters to calculate a best estimate of position based upon the data that it has available. By using an INS the position of the UAV (and its sensors) can be calculated at a higher frequency than by GPS alone; accounts for the motion of the UAV, and through applying statistical filters is able to provide a better estimate of position should information from one of the input sensors be less than optimal. In addition, by recording the platform’s (and hence, imaging sensor’s) trajectory throughout its mission, it means that there is a record of position that can post-processed and refined to provide even better positioning data for the data collected during that flight. It is this integration of positioning data that illustrates why when taking account of the movement the UAV to collect data for survey purposes, that information only from an IMU or GNSS receiver is not sufficient.
So why do you really need a survey-grade inertial navigation system?
The rise of UAV technology has led to a vast range of UAV platforms now being made available. These range from consumer to professional -grade systems, have a wide spectrum of price-points, and a wide variety of functional capabilities of components on these systems; including the autopilot and flight operations systems employed. Due to the types of materials and technologies used in professional 3D imaging systems such as LiDAR payloads, it also means that the monetary value of the imaging sensor can often be anywhere from 2-6 times the value of the UAV platform itself.
One principle of any survey-related methodology is that of consistency. Due to the variety of platforms available, it means that the quality and capability of components used on the flight control systems of any two UAV platforms can be very different to each other. The navigation capabilities of a flight control system have also often been designed to perform to different degrees of spatial tolerance than those of a system that has been designed for survey control. To give an example, data collected by a moving geospatial data collection vehicle (for example a UAV or mobile mapping vehicle) typically needs to be positioned to within +/- 5cm for it to be considered ‘survey-grade’. The autopilot system on an octocopter UAV that is often used in professional operations to carry an imaging payload will typically have a hover-positioning specification of around +/- 0.5m-1.5m. A survey-grade INS such as the OxTS xNAV550 Inertial+GNSS system calculate the spatial trajectory of a UAV platform to within +/- 2cm.
In an era when UAV practitioners need to be able to mix-and-match their payloads, it is understandable why the requirements for carrying a payload for survey applications can be different from using that same UAV platform for other purposes. By using a survey-grade INS on projects that do require it though means that the practitioner is utilizing technology that enables them to be more productive, but is also employing a validated method of positioning the data that they collect that meets the finer tolerance levels required by survey-related projects.