Previous Abstract Return to Session B6 Next Abstract

Session B6: Aviation and Aeronautics

Navigation for UAVs in Harsh GNSS Environments Using a High-Performance IMU
Kirstin Schauble, Walter Stockwell, Mike Horton, ANELLO Photonics
Alternate Number 1

The past five years have seen a revolution in small unmanned aerial vehicle (UAV) capability. Companies now routinely provide commercially available UAVs that can carry cameras, sensors, and other payloads reliably over distances of 7 to 10 km. The pilot can navigate much beyond the visual range of the pilot, but the availability of real-time positioning from the GNSS receiver is the key to these operations. UAV technology currently has capability gaps in navigation for operations in GNSS-contested environments. Not only is GNSS affected in contested environments by spoofing and high-power RF jamming, but also by obstructions such as canopy, tall buildings, deep canyons, polar areas, etc. Without a navigation solution in these GNSS-challenged scenarios, the pilot cannot reliably navigate the UAV beyond visual-line-of sight.
Previous works have studied using GBAS [1] and PPP [2] to augment or improve GPS signals for UAVs, but GPS signals themselves may not always be available or reliable. In the absence of GNSS, an inertial navigation system (INS) with gyroscopes and accelerometers must be used to maintain a navigation solution. The sensors typically used on UAVs suffer from large drifts over short time periods, which would make GNSS-denied navigation over any significant time period impossible. A low-drift gyroscope such as fiber-optic gyro (FOG) or ring-laser gyro (RLG) is needed, but these sensors have notoriously large size, weight, and power consumption (SWaP). In addition, to compensate for the accelerometer drift, a method for detecting actual airspeed is needed to keep track of distance traveled without GNSS.
In this work, we develop a silicon-based optical gyroscope (SiPhOG), which brings optical gyro performance to the mass market with a much lower cost and SWaP comparable to MEMS-based gyroscopes. Each SiPhOG has a single axis high-performance, low-drift, and low-noise optical gyro and a 3-axis MEMS IMU. We then create a 9-axis IMU with three SiPhOGs, triple-redundant MEMS IMUs for acceleration measurements, and a 3-axis magnetometer. These sensors and the CPU are combined in a small form factor to create a high-performance, low SWaP system we call the X3. The X3 is designed for easy integration with commercially available UAV flight controllers and can track angular rates with an accuracy that is not possible with MEMS-based IMUs.
Navigation without GNSS requires an estimate of distance traveled as well as heading. For distance traveled, we integrate a cutting-edge airspeed sensor that uses ultrasonic waves to measure airspeed direction around 360 degrees. An estimate of ground track or distance traveled can then be made using this airspeed sensor and the three 3-axis accelerometers in the X3.
We then integrate the X3 with a UAS using a Pixhawk controller running PX4 software. Pixhawk is one of the world's most popular hardware open-source standards for UAV flight controllers, and PX4 is the corresponding open-source software package for controlling UAVs. In this work, we modify the PX4 code to take advantage of the high-performance gyroscopes, and add sensor fusion capable of creating a navigation solution without GNSS. The sensor fusion algorithm determines the accuracy of GNSS measurements using various metrics including horizontal and vertical accuracy, number of satellites, and position dilution of precision (PDOP). It uses GNSS corrections when available, but also maintains an accurate position solution when GNSS measurements are unreliable or unavailable, which is not possible in current PX4 implementations.
There are several mission profiles we consider which would require such functionality from a UAV. In each scenario, a pilot is flying a UAV beyond visual line-of-sight using a ground control station (GCS). The pilot needs to fly the UAV to a distant location, perform a task at the location, and then return the UAV to the take-off point. GNSS would be available during normal operation, and the UAV would be reporting its position to the GCS over the telemetry link. The pilot would fly the UAV to the target location using reported GNSS position and a camera to see the target location.
In the first scenario, the UAV starts in a location where GNSS is available, then GNSS signal is lost during the flight. The UAV can initialize and launch using standard procedures, and in the beginning of the mission, the Pixhawk system is using GNSS as part of a sensor fusion algorithm to perform tasks such as estimating winds and sensor biases. At some point during the mission, the UAV enters an area where GNSS is unavailable or unreliable. The UAV automatically switches to a GNSS-denied navigation solution and the Pixhawk uses the acceleration, angular rate, and compass data from the X3, combined with information from the wind speed sensor, to estimate a position solution. The GCS will continue to show a position and heading in the absence of GNSS. When the UAV is near the target location, the pilot uses a camera to locate the target position and fly the UAV to destination. After performing the task, the pilot turns the UAV 180 degrees to fly the UAV back towards the take-off point, and any previously estimated wind conditions are used to estimate the time to fly back to the take-off point. During the return flight, the UAV will either regain GNSS fix or return to the pilot’s field of view, at which point the pilot flies the UAV back to the take-off point per normal operation.
In the second scenario, GNSS is unavailable during the start of the mission but becomes available during some portions of flight. In this case, the Pixhawk system needs to be initialized with an estimated location and heading for the UAV using the GCS. The pilot launches the UAV, and the Pixhawk uses the X3 acceleration and rate sensor data, combined with the wind speed data and compass to estimate an updated position solution. The GNSS receiver will always be trying to measure a position fix, and at any point during the flight, if the signals are of sufficient quality for long enough to obtain a fix, the receiver would output a position solution. The updated GNSS position is then used as part of the sensor fusion algorithm to estimate winds or sensor biases. As in the first scenario, when the pilot locates the target position, they fly to the destination, perform the task, the turn the UAV 180 degrees to reverse course. The pilot uses the known GNSS positions when available to understand how the GNSS-denied positions might be affected by wind and use this information to fly the UAV back to the take-off location.
These and many other mission profiles would not be possible with the existing Pixhawk sensors and PX4 software. In this presentation, we will demonstrate that these mission profiles are possible using the X3 and updated PX4 initialization and GNSS-denied sensor fusion algorithm. We will do a deep dive into the gyroscope specifications required to complete the three missions under a variety of harsh GNSS-denied scenarios. We will then uncover the device physics that enables the high performance of the X3 with a form-factor, weight, and power consumption low enough to be suitable for UAVs. We will also describe the airspeed sensor performance requirements needed to accompany the X3 in accomplishing these missions, and show our results from the airspeed sensor used in this work. This presentation will glean important insights toward removing the dependence on GNSS and enabling high accuracy navigation for beyond visual line of sight UAV flight missions.
[1] Felux, Michael, Jochems, Sophie, Schnüriger, Philipp, Fischer, Valentin, Steiner, Patrick, Jäger, Michael, Sarperi, Luciano, Cacciopoli, Natali, "Flight Testing GBAS for UAV Operations," Proceedings of the 35th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2022), Denver, Colorado, September 2022, pp. 1576-1588.
[2] Yamada, Hideki, Matsushita, Saya, Kawano, Isao, Inoue, Koichi, Takasu, Tomoji, "Mitigation of the Phase Wind-up Effect for PPP Ambiguity Resolution using INS and Geometryfree Approach in UAV Experiments," Proceedings of the 34th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2021), St. Louis, Missouri, September 2021, pp. 4170-4182.

Previous Abstract Return to Session B6 Next Abstract