SIVAQ. Signal Integrity Verifying Autonomous Quadrotor

Transcription

1 SIVAQ Signal Integrity Verifying Autonomous Quadrotor

2 The Team 1 Brett Wiesman Nick Brennan Steve Gentile Ross Hillery Shane Meikle Erin Overcash Sean Rivera Geoff Sissom Matt Zhu

3 Agenda 2 Objectives Project description CONOPS FBD Requirements Feasibility Studies Further Testing Questions

4 Project Description 3 Mission Statement: Augment the capabilities of the Parrot AR Drone 2.0 such that it flies autonomously with a predetermined flight path, records data, relays data, and detects and responds to GPS Radio Frequency Interference (RFI). Mission Objectives: Establish a system that can detect GPS signal interference Install extended range communication hardware on the drone Develop software for communication, data processing, and piloting of the drone by selecting waypoints on a map at a ground station [1].

5 Project Description 4 Highest Level of Success [2] : Autonomous quad rotor autopilot with: (a) (b) (c) (d) (e) (f) (g) (h) (i) GPS navigation system and signal integrity monitoring; Return home" capability; Upgraded battery; Custom housing and mechanical interface between electronics and vehicle; Extended-range, 2.4GHz Wi-Fi communications device for transmission of video, data, and last known position. During 1 minute loiter SIVAQ will provide live video data, such that the pilot can identify a red target 1 m 2 in a 3600 m 2 field. 2D mapping of RFI area and photography to attempt to locate the RFI source; Interface for future additional sensor package; Custom fuselage that improves efficiency while preserving center of gravity and structural integrity and ensures control algorithms will maintain stability. New fuselage will house (a), (c), (e), and (h).

6 Baseline Vehicle Hardware [3] 5 USB port 400 Mb/s Battery Lithium Polymer 1000 mah capacity Microprocessor 16 bit PIC 40 MHz clock Motor (x4) Brushless 14.5 Watts 28,500 RPM Microball Berings Nylatron Gears Memory DDcaR2 RAM 200 MHz clock Motor Controller (x4) Microprocessor 32 bit ARM Cortex A8 1 GHz clock Downward facing camera QVGA 64 diagonal lens 60 fps recording speed Wi-Fi Atheros AR61036 chipset 2.4 GHz Tx frequency Motherboard Forward facing camera 93 wide angle lens 720p 30fps recording speed Pin Connection Solder Connection Navboard Altimeter Barometric pressure sensor 10 Pa precision Ultrasound 6 m precision Accelerometer 3 axis 50 mg precision Magnetometer 3 axis 6 precision IMU Invensense IMU-3000 Contains 3 axis gyro and input for 3-axis accelerometer

7 Functional Black Diagram 6 GPS Antenna GPS Receiver Storage Device Thermistor RFI Simulation Autonomous Navigation Software RFI Detection Software Battery Vehicle Modifications USB Hub/Port Electronics Package Vehicle Kill Command Dynamic Waypoints Navigation Software USB port Wi-Fi Motherboard GUI Wi-Fi AR Drone 2.0 Hard Data/Power Connection Wireless Data Connection Developed for Project Pre-Existing Hardware

8 CONOPS 7 Begin flight with continuous signal integrity monitoring and flight data transmission Command Destination and Waypoints for autonomous travel Travel towards estimated target location Return home Loiter 1 minute and locate target using downward facing camera Downlink and store flight data in real time

9 CONOPS Scenario 2 8 Continuous signal monitoring and data transmission detection Immediate, large radius RFI is enabled Command Destination and Waypoints for autonomous travel Immediate, powerful RFI detected! Lose Communication link with ground station. Downlink and store flight data in real time Abort mission, disable GPS and attempt to return home inertially

10 CONOPS Scenario 3 9 Continuous signal monitoring and data transmission detection Command Destination and Waypoints for autonomous travel False GPS sphere of influence False Signal Detected! Viable communication link with ground station. Downlink and store flight data in real time Attempt to map sphere of influence or locate source of RFI

11 Functional Requirements [1] 10 REFERENCE Description 1.QUADFR.1 SIVAQ shall travel autonomously via predetermined waypoints while maintaining pseudo range accuracy of 7.8 meters (TBR) at 95% confidence level. 1.QUADFR.2 1.QUADFR.3 SIVAQ shall monitor GPS information integrity and detect radio frequency interference. Signal shall be considered compromised if AGC level is greater than three from nominal AGC level. SIVAQ shall create map RFI zone of influence. 1.QUADFR.4 1.QUADFR.5 1.QUADFR.6 1.QUADFR.7 1.QUADFR.8 1.QUADFR.9 1.GRNFR.1 1.GRNFR.2 SIVAQ shall return to ground station once mission is completed or at 30% (TBR) battery life remaining. SIVAQ shall be able to fly to a target location within a 3km radius area, capture video data for 1 min, then return home. Finalized SIVAQ cost shall be less than $750 (TBR) in components. SIVAQ shall be operational in open terrain and in conditions of ideal weather (no precipitation and no wind). SIVAQ shall be equipped with an extended range 2.4 GHz Wi-Fi two-way communications device. During 1 minute loiter SIVAQ will provide live video data, such that the pilot can identify a red target 1 m 2 in a m 2 field. A GUI will allow the user to select waypoints and specify the vehicles mission. Additionally, the user will be able to alter waypoints during flight. The command center must be able to receive and display data sent from the vehicle for processing.

12 Baseline Design Selections [1] 11 CDD Design Solutions Subsystem RFI detection Mass and Power Communications Mission Data Control Software Navigation method GUI Software Design Selection Automatic Gain Control signal monitoring Structural modification and high capacity Li-Ion battery 2.4 GHz Wi-Fi antenna modification USB flash storage and Wi-Fi n streaming AutoPylot written in MATLAB GPS based waypoint navigation NASA WorldWind mapping software and Java Swing

13 Feasibility Studies 12 Aspects of the project analyzed for feasibility: 1. How will the vehicle detect and simulate RFI? 2. Is 6 km range (3km out and back) possible given mass and power considerations? 3. Can the vehicle maintain communication link with the ground station over 3km distance? 4. Can vehicle accommodate required data transfer and storage rates? 5. Does MATLAB have an acceptable response time for control software? 6. Is waypoint style navigation possible with selected software?

14 RFI Detection Feasibility 13 Automatic Gain Control Automatic Gain Control (AGC) is an adaptive circuit which dynamically adjusts incoming signal gains to match the level requirement of downstream electronics [4] AGC is low complexity and is intrinsic in all multi-bit GPS receivers AGC is not part of NMEA message but some civilian GPS receivers include it in the digital output GPS RFI can be detected by monitoring the output of the AGC circuitry A boosted GPS signal causes the AGC to drop

15 RFI Detection Feasibility 14 A study conducted in Sweden [4] using a GPS RFI device showed that AGC can be used to accurately detect GPS RFI AGC output is monitored for a drop below a predetermined threshold This proves the feasibility of the AGC detection method

16 RFI Detection Feasibility 15 AGC Signal Characterization AGC levels are subject to noise from surroundings [3] Refinements can be made to AGC measurements by accounting for thermal noise and other signals present in the frequency band Noise can be characterized for the specific GPS module and flight conditions to properly detect active RFI

17 RFI Detection Feasibility 16 Thermistor RFI Ground Testing - Noise Characterization Ground and Flight models of the GPS modules will be bench tested and data will be logged using a data acquisition system Tests will be run both on and off the UAV platform On-platform tests will be run with the vehicle off, on and in controlled flight Results from this test will allow for the characterization of noise in the GPS signal for various conditions

18 RFI Detection Feasibility 17 Thermistor RFI Ground Testing - L1 Injection An L1 GPS signal simulator will be fed through a variable gain amplifier and injected into the GPS RF stream, simulating a malicious RFI source A data acquisition system will be used to monitor and log the AGC signal from the GPS module This ground test will be used to characterize AGC response to an amplified signal in the L1 band and determine the appropriate threshold at which AGC will accurately denote GPS RFI

19 RFI Simulation Feasibility Nav Software 18 Thermistor RFI Software Waypoint-based RFI Simulation To implement RFI in flight (currently illegal), data from RFI ground tests are used to create a software simulation which mimics the results of actual RFI on the AGC signal in the GPS module s digital output GPS waypoints are used to define the perimeter of the simulated RFI zone which, when breached, will result in simulated RFI; manipulating the digital AGC signal

20 Range Feasibility Analysis 19 Customer Requirement: AR.Drone 2.0 shall fly 3 km from launch point, loiter for 60 seconds, then return 3 km to takeoff point. AR.Drone 2.0 Capabilities: Manufacturer Claim: 3.6 km max range (Cruise Speed 5 m/s, Flight Time 12 minutes) User Claim: 2.25 km max range (Chuck Rossetti claims only Wi-Fi modifications) Drivers for Range Requirement: Commercially available radio frequency interference devices have extremely variable areas of influence (up to 6km) but the typical RFI device zone of influence extends 2 to 500 meters from signal source. Extended range improves functionality of AR.Drone 2.0, enables ability to map zone of GPS corruption. Range improvement enables increased flexibility in future missions (Customer Request).

21 AR. Drone Mass Budget 20 Required Components Component Mass [g] Percent of Stock Mass [%] Stock Outdoor Vehicle GPS Receiver/Antenna 4000 mah Battery (Dynamite Speed Pack Silver) Polarized Antenna USB Magnetic Compass Total Mass Unrequired Components Component Mass [g] Percent of Stock Mass [%] Outdoor Hull 1000 mah Battery Stickers USB Port Camera Arm Total Mass Final Mass [g] [% of stock]

22 Range Feasibility Analysis 21 Estimating Current Draw STEP 1: Find current during hover STEP 2: Find flight angle at designated speed Amps hover A Velocity Battery Pack Thrust Weight Thrust Angle Weight

23 Range Feasibility Analysis 22 Power Draw Estimation for 6 km distance and 3600 m 2 scan Component Current [Amps] Battery Capacity [mah] Percent of 4000 mah Battery [%] GPS Receiver/Antenna USB Flight* Scan* Battery Margin NA Total *Stock hardware, processing, and telemetry of A.R. Drone 2.0 is included in flight and scan The necessary battery capacity to complete the mission is larger than the heaviest allowed battery

24 Battery Mass [g] Range Feasibility Analysis 23 Battery Mass to Capacity Ratio Study 250 Design Area Unfeasible Area Needed to fulfill 6km requirement Capacity [mah]

25 Communication 24 Requirements Vehicle must continuously transmit live video data to the ground station A pilot at the ground station must be able to send vehicle kill command (FAA Requirement) [5] User must be able to dynamically change waypoints mid-flight Vehicle is equipped to communicate via Wi-Fi Average Wi-Fi range over protocol is when outdoors This will not reach the suggested travel distance of 3 km The communication capability must be upgraded in some way in order to maintain a high quality data link at the suggested range

26 Communication Modification 25 Antenna Modification Baseline communication range calculation Normal Wi-Fi range 100m Signal strength drops with range squared so for each doubling of range we need 6dB more signal gain, the range must be doubled 5 times (30dB gain) in order to achieve 3km distance assuming no transmission loss Further testing will be necessary to obtain transmission data from the vehicles antenna Must also consider FCC regulations and may not exceed 1W (30dBm) TRP, 4W (36dBm) EIRP* Highest gain commercially available omnidirectional antenna found was only 15 db *For point-to-point link, antenna gain can be increased to obtain EIRP > 36dBm but for every 3dBi increase in antenna gain, transmit power must be decreased 1dBm 3km range cannot be obtained legally using omnidirectional antenna Online user Garrock has created and provides detailed descriptions of what he labels a Wheel Antenna Mod [6] which modifies the vehicles built in antenna to increase the transmission capability to a range of 200m Several users have successfully used the Itelite SRA A 19 dbi directional gain antenna on the ground station to increase the range of the vehicle to a distance of 1 km [7]

27 Data Transfer and Storage 26 Requirement: Data Storage GPS integrity information, location, velocity, heading, IMU and other sensor data Data Streaming Constant real time video Kill command GPS integrity information, location, velocity, heading, IMU and other sensor data GPS Integrity Info

28 Data Transmission Feasibility 27 Data Transmission Rates of On-Board Electronics* Data Rates Invensense IMU 3000 Front Facing Camera Down Facing Camera GPS Battery Levels Needed Data Rate Atheros AR Capability [3] 2 MB/s 0.5 MB/s MB/s Ω MB/s 1 MB/s 3.74 MB/s 1 MB/s 9 MB/s * Values used in table represent upper limit of data rates found in research Functions of sampling rate (IMU sampling at 32 ms) Ω Assumed to be 1/3 of front facing camera because resolution is 120p at 60 fps vs. front facing camera s 720p at 30 fps. Further testing is required to verify resolution Average used as Wi-Fi data sheet reads. Further testing required to verify. The vehicle will be able to transmit all necessary data to the ground station. If Wi-Fi power drops below threshold, streamed info will be buffered but still come through

29 Data Storage Feasibility 28 Verbatim Tuff- N -Tiny USB Drive only weighs 1 g [8], by far the lighted storage device found to meet project storage needs Required Data Rate Required Storage Capacity Data Storage for On-Board Electronics Verbatim Tuff- N -Tiny USB Drive Write Speed Verbatim Tuff- N -Tiny USB Drive Read Speed Verbatim Tuff- N -Tiny USB Drive Capacity MB/s* 4 GB* 14 MB/s 27 MB/s 16 GB * Data rate excludes video not being stored **Assuming a 20 minute flight Additionally, the onboard ARM Cortex A8 is capable of writing at 20 MB/s over USB [12] USB Drive can store approximately 80 min of flight data 4 x faster than the necessary data rate

30 Onboard Software 29 Requirement: On-board software must allow for autonomous waypoint navigation Must integrate the AR Drone s control algorithms Team familiarity more important than performance AutoPylot Control Software: Free/Open Source C, Python, and MATLAB integration Runs using AR Drone s control algorithms Language Performance (MIPS) Threading Estimated Response Time [ms/mega-instruction] C 2000 Yes.5 Python 400 [9] No 2.5 MATLAB* 300 [2] Yes 3.33 MATLAB compiled to binary using MEx to run on vehicle AutoPylot with MATLAB chosen for familiarity with language has acceptable response delay

31 Autonomous Waypoint Navigation 30 AutoPylot [13] Developed for 64-bit Linux Compiles AR.Drone 2.0 SDK using latest vehicle firmware Preserves Parrot calibrated control algorithms Capable of working with with MATLAB using Mex to compile binaries

32 Autonomous Waypoint Navigation 31 W U + - e D G + + Y Variable U e D G W Y Description Destination (Lat, Lon, Alt) Error between current and desired position AutoPylot control commands (zap, phi, theta, gaz, psi) Vehicle built in dynamics and flight control External disturbances altering position (wind) Current location (Lat, Lon, Alt) Parrot currently manufactures and sells its own GPS module, thus the latest vehicle firmware is already capable of interpreting GPS signal data AutoPylot uses a commands matrix that it sends to the vehicles onboard processing unit The commands matrix contains values for roll, pitch, yaw, zap, and gaz that move the vehicle A transformation that creates these commands from GPS position will need to be developed Waypoint navigation with the AR. Drone 2.0 has been successfully completed by several users and institutions including the Delft University of Technology [10] Waypoint Navigation is Feasible.

33 GUI Software 32 The GUI must: Allow user to select waypoints by clicking on a map. These waypoints are then communicated to the vehicle real-time and can be changed mid-flight Display real time video data streamed from the vehicle Enable quick use of a kill command Allow user to monitor telemetry data (battery life, location, velocity, connectivity, GPS integrity) NASA s WorldWind was selected for API support and team familiarity with Java Alternative is currently Marble (uses Python)

34 GUI Mockup 33

35 Further Feasibility Studies 34 Determine if navigation with vehicle s IMU and other sensors bring vehicle home within acceptable error without the aid of GPS Determine most efficient RFI mapping method

36 Navigation Without GPS 35 Customer Requirement: AR.Drone 2.0 shall be able to return home when GPS signal integrity is compromised using a secondary navigation system. Solution: The AR.Drone s velocity measurements will be integrated in both horizontal axes to calculate the AR.Drone s position. This will then be used for navigation back to the home base. V y Last Known GPS: Lat: N Lon: W X: 0 m Y: 0 m V x Known GPS: Lat: N Lon: W X: 1243 M Y: 345 M

37 Speed [m/s] Inertial Navigation Testing 36 Method: AR Drone calculates velocity by integrating accelerometer measurements, corrected by downfacing camera. Precision of accelerometers is added to the steady accuracy of the camera to provide a velocity solution that does not compound error, and a position solution that compounds linearly rather than exponentially. Velocity Estimation Time [s] [15] Assuming 3 km path traveled at 4 m/s and 0.5 m/s error, distance error would accumulate to 375 m Test Method: Flight test logging GPS and IMU data and compare

38 RFI zone of influence mapping Mapping technique is going to depend highly upon AGC lab measurements as well as onboard processor availability Lab testing will need to determine nominal AGC value as well as necessary change that corresponds to RFI Once this is known appropriate mapping algorithm will be developed In addition to mapping technique, minimizing total distance traveled without sacrificing map accuracy is desired 37

39 RFI Mapping preliminary CONOPS d 1 AGC value again trusted, store latitude distance between good locations d 2 D2 < D1,Mapping completed, return home AGC threshold crossed! Store last known good location AGC level returned above threshold! Store location

40 Feasibility Summary 39 Subsystem Feasible Reason RFI Yes Previous testing demonstrated feasibility. Mass and Power No MATLAB model showed unfeasibility with commercially available batteries. Communications No Cannot reach required range legally using 2.4GHz Wi-Fi. Storage and Data Yes Worst case analysis still within bandwidth. Software Yes Delay using MATLAB is less then human noticeable delay. Navigation Yes Previous work with AR Drone 2.0 demonstrated waypoint travel. GUI Software Yes NASA WorldWind capable of interfacing with Java 3km range requirement will need to be reassessed with customer

41 Conclusion 40 Areas of concern 3km range requirement is large, affects power, mass, and communication feasibility and the reason for its existence is unclear Inertial sensor navigation introduces large error and will require extensive testing to understand return to home feasibility Overall Project feasibility The proposed project will be feasible if the range requirement can be reduced

42 References 41 [1] Brennan, Gentile, Hillery, Miekle, Overcash, Rivera, Sissom, Wiesman, Zhu, SIVAQ Conceptual Design Document, University of Colorado Department of Aerospace Engineering, 30SEP2013. [2] Brennan, Gentile, Hillery, Miekle, Overcash, Rivera, Sissom, Wiesman, Zhu, SIVAQ Project Definition Document, University of Colorado Department of Aerospace Engineering, 23SEP2013. [3] Technical Specifications: State of the Art Technology, Parrot AR Drone 2.0, [ 2/specifications/] [4] Akos, D. M, Who s afraid of the spoofer? GPS/GNSS spoofing detection via automatic gain control (AGC), Navigation, 59(4): , [5] COA Notes, 19SEP2013, [ [6] Garrock, Wheel Antenna Mod Significant Wifi Performance Upgrade, Parrot AR Drone & AR Drone 2.0 Forum, 18JUL2012, [ [7] Parrot AR Drone & AR Drone 2.0 Forum, [ [8] Verbatim USB Storage Website. 16GB - TUFF-'N'-TINY USB Drive, Copyright 2013 [ [9] PyPy Speed Center, [ [10] Full Blown Hucker, TU Delft Search and Rescue with AR Drone 2, MultiRotorForums.com, OCT2012, [ 2&s=f93bdfa922cee f08fe267be68%00] [11] Bristeau, P. J., Callou, F., Vissière, D., Peiti, N., The Navigation and Control technology inside the AR.Drone micro UAV, [ [12] Texas Instruments, AM335x-PSP Features and Performance Guide, [ [13] Levys, AR.Drone AutoPylot, [