GNSS for UAV Navigation Sandy Kennedy Nov.15, 2016 ITSNT
Sounds Easy Enough Probably clear open sky conditions?» Maybe not on take off and landing Straight and level flight?» Not a valid assumption for all applications How does GNSS fit into the UAV navigation system? 2
UAV Navigation: Control System Telemetry GNSS IMU? GNSS Receiver Host Processor (Flight Control) Actuation of Vehicle Controls Telemetry Rx/Tx 3
Requirements Depend on the Application What is the required GNSS positioning accuracy?» Just metres?» Centimetres?» For the whole flight? For landing and take off? What is the allowed envelope for GNSS antenna and receiver?» Antenna mounting location» Antenna size and weight» Same for receiver What is the operating environment?» Actual open sky?» Rotor blades?» Buildings/structures?» Launch dynamics, and expectation of accuracy availability 4
Antennas the Neglected Front Door Positioning accuracy drives antenna specification for phase centre stability and uniform gain pattern» Across frequencies and across elevation/azimuth angles» Becomes more challenging with smaller size, but size constraint is typical» Antenna position is usually not what is required. Needs to be translated to a location of interest (underbelly/wing hook, landing gear) Frequency support» Which GNSS signals to support» L-band corrections?» Multipath rejection performance? Mounting constraints» under rotor blades» 5 changing orientation from take off to flight
Point of Interest vs Antenna Mounting Location 6
Changing Orientation 7
Antenna Size/Space Constraint 8
Telemetry Some other form of RF onboard the UAV» For command/control of the UAV» For delivery of GNSS differential correction data» UAV reporting back to base Needs an antenna too Potential interference source» Separating antennas may not be possible on small platforms 9
IMU Provides the attitude information needed to translate antenna position Full GNSS/INS solution or just AHRS?» AHRS typically assumes flat and level and constant velocity operation» Position bridging needed or just attitude? May be processed by GNSS receiver or the host processor. Needs GNSS data to estimate/calibrate the IMU aka converge the GNSS/INS filter. 10
GNSS Receiver Computes position and velocity» May also compute GNSS/INS solution» Could provide the input to another sensor fusion filter» Application specifics dictate what type of solution is needed Epoch to epoch as uncorrelated as possible, or As smooth as possible to feed the control system» SWaP constraints one or all May be receiving correction data» L-band for PPP» Differential corrections for RTK» Differential corrections for moving base (relative RTK) 11
Landing Scenarios 12
Host Processor(s) Interfaces with the GNSS receiver(s), and possibly other sensors» A backup GNSS receiver is sometimes used Performs flight control computations Actuates the vehicle controls Manages communications with the user/base Potential source of interference» CPU clock frequency creating harmonics in-band 13
Payload Could be inert» Packages» Medical products Could be more electronics»lidar» SAR» Cameras Potential source of interference 14
Blood Product Delivery 15
Synthetic Aperture Rader 16
Common Theme? Lots of potential for self-interference» Electronics closely packed together = self-interference How much degradation is acceptable in your design? Depends on performance requirements and operating conditions How much margin do you need to leave for dealing with external interference?» Autonomous flight means your navigation must be reliable, with maximum availability» Risk management standards higher 17
How to Deal with Self-Interference First step is to detect and then diagnose the source» Low C/Nos» Frequent lock breaks» Increased phase noise» Slow time to first fix» Difficulty fixing ambiguities RF spectrum analyzers are complicated to deploy properly, and interpreting their measurements can be difficult Could use mechanical solutions» Shield the GNSS receiver card (metal can) Won t help if the antenna is the interference entry point» Separate GNSS and telemetry antennas Not always possible due to size constraints There s got to be a better way 18
Turn your GNSS Receiver into a Spectrum Analyzer Build the detection capability into the GNSS receiver, as we have in OEM7» Gao, F., Kennedy, S., (2016). Demonstrated Interference Detection and Mitigation with a Multi-frequency High Precision Receiver, Proceeding of ION GNSS+ 2016, Portland, Oregon Spectrum Analysis via FFT» Ideal for in-band Continuous Wave (CW) interferers which are common in self interference cases host processor CPU clock harmonics» More sophisticated shape analysis can determine out-of-band effects Other detection methods within the receiver signal processing chain» Borio, D. et al (2016) Impact and Detection of GNSS Jammers on Consumer GradeSatellite Navigation Receivers, Proceedings of the IEEE, Vol. 104, No.6. Run built in tests that continuously monitor for the presence of interference, and report to the user 19
Spectral Analysis via FFT: 3 options to output ADC Pre-decimation: full path view, good for detecting near band interferers lurking in the transition zone Post-decimation: the bandwidth of interest for a particular signal (ie GPS L1) Post-filter: the band of interest for a particular signal after the application of a notch or bandpass filter 20
Normal GPS L1 Power Spectrum 21
Mitigated In Band CW Interferer with Notch Filter 22
Out of Band Wideband Interferer 23
Interference Confirmed Now to Mitigate Interference Toolkit is onboard every OEM7, and is SW configurable on or off HDR Mode:» High resolution ADC More range to the upper end to handle strong interference signals» Increased front end linearity Pass band as flat as possible» Multi-stage gain control Distribute gain to keep signals in the middle of device s operational range Digital Filtering» Notch and bandpass filters 24
Example: Interference Toolkit vs Cavity Filter Out of Band Narrowband Interferer: GlobalStar Satellite voice/data service 25
Viewing the Spectrum 26
GPS Satellites Tracked without Mitigation 27
Single Point Pseudorange Position No Mitigation 28
GPS Satellites Tracked with Mitigation 29
Single Point Pseudorange Position with Mitigation 30
Positioning Errors Before and After Mitigation No Mitigation Northing Easting Height Std.Dev. Range (m) Std.Dev. Range (m) Std.Dev. Range (m) (m) (m) (m) OEM628 3.364 126.494 1.164 34.388 5.783 230.399 OEM719 1.420 36.663 1.155 34.350 1.242 32.655 With Mitigation Northing Easting Height Std.Dev. Range (m) Std.Dev. Range (m) Std.Dev. Range (m) (m) (m) (m) OEM628 0.122 0.566 0.113 0.460 0.291 1.283 OEM719 0.129 0.679 0.114 0.522 0.260 1.231 31
Multi-Frequency, Multi-Constellation Approach GPS only is not sufficient for most applications with restricted sky view, or high banking turns L1 only doesn t provide sufficient accuracy and time to fix over longer baselines So, expand to track everything in the sky.» Won t diversity of measurements help overcome interference? Yes, it will, to a certain extent» Provided the interference is restricted to a single frequency or signal type» Provided the interference is at a low enough power that the front end doesn t saturate 32
Example: In-band CW Interferer Placed an CW interferer at 1582.5 MHz, -44 dbm strength, via conducted input at the receiver input» Mitigated with HDR mode and a notch filter, 1 MHz wide centred at 1582.5 Computed an RTK solution over a very short baseline (5m), using a clean base receiver, without interference Processed the collected measurements in three ways:» L1 only (GPS, GLO, GAL)» GPS L1/L2» All in view (GPS L1/L2/L5, GLO L1/L2, GAL E1/E5AltBOC) 33
Available Satellites 34
In Band Continuous Wave Interferer 35
Signals Tracked 36
L1 Only 37
GPS L1/L2/L5 38
All in View: GPS L1/L2/L5, GLO L1/L2, GAL E1/E5 39
Summary of RTK 3D Position Errors 40
Summary GNSS for UAV Navigation Protect your GNSS solution Backups are good, but exploit GNSS to it s full extent. Will only help your sensor fusion since GNSS is typically used to estimate errors in other sensors 41
Thanks for your attention!
Receiver Design for Interference Robustness Filters ADC ADC Baseband Processing 43 Variable Gain Amplifiers
OEM7 Family OEM7600 55x35mm OEM719 OEM7700 OEM7720 71x46mm OEM729 100x60mm 44