EarTh HOrizon Sensor Critical Design Review Team Noah Buchanan Matthew Busby Matthew Cirbo Taylor Dean Jesse Keefer Patrick Klein Thomas Konnert Cole Oppliger Neal Stolz 7/12/2016 Customers Joe Breno Randy Owen Advisor Dr. John Farnsworth University of Colorado Aerospace Engineering Sciences 1 Outline • • • • • • • Outline 7/12/2016 Purpose Design Solution Critical Project Elements Requirements and Satisfaction Risks Verification and Validation Project Planning Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 2 Purpose • Sun sensors are inaccurate • Star trackers are too expensive • Surrey’s customers desire a Goldilocks solution – – Not too expensive Accuracies between that of sun sensors and that of a star tracker Depiction of satellites in low Earth orbit [1] Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 3 CONOPS Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 4 CONOPS Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 5 Mounting Angle Customer is responsible Horizon Not for mounting ETHOS to Visible the satellite at the correct angle θ1 ≠ θ2 Horizon Visible Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 6 Solution: ETHOS • Customer Requirements - DR.1.1.2: Perform During Simulated Eclipse Period DR.3.1.1: Provide displacement at 10 Hz DR.3.5: Accept Voltage between 22 & 34 V - Power Regulator Board BeagleBone Black • Solution - Outline 7/12/2016 BeagleBone Black Custom Power Reg. Board FLIR Tau 2 IR Camera Solution CPEs Camera Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 7 Solution: ETHOS • Customer Requirements – – • DR.3.2: housing built less than or equal to: 4.21”x3.74”x2.48” DR.3.3: Mass less than 600g Solution – – Built out of Aluminum 6061 Mass of Housing ~412g 2.48” 4.21” Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences 3.74” Algorithm V&V Planning 8 Solution: ETHOS • Minimize mass and Maximize Volume – – 1/8” Aluminum (Left, Right, Top, & Bottom) Machined 3/8” Aluminum (Front & Back) #6 Screws 3.49” 1.18” 2.23” 0.125” .125” Thickness #3 Bolts Outline 7/12/2016 Solution CPEs Camera Centered on Face, Diameter 1.142” 0.31” 0.375” Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 9 Critical Project Elements • Camera Sensor and Lens • Image courtesy beagleboard.org Image courtesy flir.com • Electrical Components • Software Algorithm Image courtesy faxo.com Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 10 Critical Project Element CAMERA Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 11 Infrared Camera Solution • Camera Selection Limitations – – – – Limited selection of 90° FOV Long Wave Infrared (LWIR) cameras Limited number of LWIR cameras below $5,000 Unfamiliar camera interfaces Subject to control by ITAR (22 CFR 121.1) • FOV vs. Infrared – – – Outline 7/12/2016 Infrared band operation more important to Surrey Satellite No longer have a FOV requirement Higher FOV still desirable for larger testing range Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 12 Infrared Camera Solution FLIR Tau2 Low Res FLIR Tau2 with 7.5 mm lens Spectral Range 7.5 – 13.5 μm Resolution 162x128 pixels FOV 63° x 50° Dimensions 1.75” x 1.75” x 1.93” Power <1.0 W Sensitivity <50 mK Weight 72 g Price $2,090 Lead Time 2-4 Weeks 1.75” 1.75” 1.93” Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 13 Camera Requirements Design Requirements Design Solution DR.1.1.1 Observe Earth outside of simulated eclipse region IR spectrum (8-14 μm) allows imaging outside and inside eclipse region DR.1.1.2 Observe Earth inside of simulated eclipse region 0.275° pixel error due to 63°x50° FOV and 162x128 pixel resolution DR.2.2.1 Displacements have errors less than 0.5° Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 14 Camera Requirements Design Requirements Outline 7/12/2016 Design Solution Camera Core Dimensions: DR.3.2.1 Volume not exceeding 4.21”x3.74”x2.48” 1.75”x1.75”x1.93” DR.3.2.2 Mass less than 600 g Mass of camera = 72 g DR.3.2.3 Power draw less than 5 W Camera draws < 1 W Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 15 Critical Project Element ELECTRICAL Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 16 Electrical Introduction Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 17 Electrical Requirements Design Requirements Design Solution DR.3.2.4 Input of 22 to 34 V 260 kHz switching voltage regulator (approx. 80% efficiency) DR.3.2.3 Power draw less than 5 W Total Power Consumption: 2.83 - 4.13 W DR.3.1 Communicate over CAN bus protocol BeagleBone Black has built-in CAN bus protocols Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 18 Electrical Requirements Design Requirements Design Solution DesignDR.3.1.1 Requirements Design Solution Need: 0.339 MIPS Have: 1000 MIPS Return displacement vector at 10-20 Hz DR.2.4.3 Store 200 minutes of telemetry data Need: Approx. 720 kB Have: Expandable - SD slot DR.3.2.1 No dimension greater than 4.21”x3.74”x2.48” BeagleBone Black: 3.40”x2.15”x0.19” Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 19 Camera – Computer Interface Pixel data Camera Generated Clock @ 10.519 MHz 1. Camera sends 8 parallel bits and clock signal Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 20 Camera – Computer Interface Pixel data Camera Generated Clock @ 10.519 MHz 1. Camera sends 8 parallel bits and clock signal 2. Signals are received via Digital I/O (DIO) pins Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 21 Camera – Computer Interface Pixel data Camera Generated Clock @ 10.519 MHz 1. Camera sends 8 parallel bits and clock signal 2. Signals are received via Digital I/O (DIO) pins 3. Built-in microcontroller pulls data from pins Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 22 Camera – Computer Interface Pixel data Camera Generated Clock @ 10.519 MHz 1. 2. 3. 4. Camera sends 8 parallel bits and clock signal Signals are received via Digital I/O (DIO) pins Built-in microcontroller pulls data from pins Data passed to processor Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 23 Electrical Component Specs Camera – Tau 2 BeagleBone Black Voltage Regulator – LM2675 Image Size 20.48 kB Processor 1 GHz AM3358 Vin 8 – 40 V Data Out 8-bit parallel @ 10.519 MHz Ram 512 MB Vout 5V Non-volatile Memory 4 GB built in & μSD expansion 1A Communication CAN bus, USB, Ethernet, & HDMI Current Out (max) Switch Frequency 260 kHz Vin 4.0 – 6.0 V PRU microcontrollers Outline 7/12/2016 Solution CPEs Camera 2x200 MHz programmable real-time units Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 24 Critical Project Element SOFTWARE Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 25 Software Requirements Design Requirements Outline 7/12/2016 Design Solution DR.2.4.1 Calculate displacement vector at 10-20 Hz Software pipelining between horizon algorithm and reading camera output DR.2.4.2 Save health telemetry at 0.5 Hz CPU runs health telemetry diagnostic routine when idle DR.3.1 Communicate over CAN bus protocol Dedicated PRU that handles CAN communication Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 26 Software Overview Reads in IR sensor data and prompts CPU to calculate displacement vector Handles CAN messages and relays commands to CPU PRU0 PRU1 CPU Runs health telemetry diagnostic routine when idle Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 27 PRU0 FBD BeagleBone Black DIO Pins IR Sensor • IR sensor outputs data to DIO pins • PRU writes data to RAM • CPU reads RAM and calculates displacement vector • CPU saves displacement vector to SD card (with timestamp and flags) Outline 7/12/2016 Solution CPEs Camera PRU0 RAM CPU SD Card Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 28 PRU1 FBD BeagleBone Black Satellite CAN bus DIO Pins PRU1 • PRU1 reads message • Saves message to buffer, if intended recipient • CPU reads message • Outputs displacement vector or health telemetry over CAN on command Outline 7/12/2016 Solution CPEs Camera RAM CPU SD Card Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 29 Timing Scheme Power-on Process Process Command Command Horizon Algorithm Initialization Diagnostics Routine CPU Horizon Algorithm Frame 1 Decode Sensor Data PRU0 Idle Frame 2 Decode Sensor Data Idle Frame 3 Frame 2 Frame 1 Decode Sensor Data Filter CAN messages PRU1 Command Command (1/Frame Rate) seconds Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 30 Critical Project Element ALGORITHM Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 31 Algorithm Requirements Outline 7/12/2016 Design Requirements Design Solution DR.2.2.2 Calculate 2-D displacement angles Circular least squares fit DR.2.2.1 Displacements have errors less than 0.5o Error correction polynomial reduces errors to ≤ 0.06o DR.2.4.1 Calculate displacement vector at 10-20 Hz C and C++ code runs at ≥ 70Hz Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 32 Algorithm Overview Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 33 Determining Start Edge • Take average of every edge through a width of B • The ‘start edge’ is the edge with the lowest average intensity • The edge search direction is perpendicular to starting edge Outline 7/12/2016 Solution CPEs Camera B B Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 34 Edge Verification • • Search for pixel with intensity greater than threshold Look beyond the pixel • If pixels beyond threshold pixel are ≥ threshold value, then it is an edge • If pixels beyond threshold pixel are < threshold value, then it is not an edge *Red edge line produced by current algorithm Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 35 Algorithm Overview Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 36 Edge Searching Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 37 Algorithm Overview Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 38 Algorithm Overview Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 39 Least Squares Fits Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 40 Algorithm Overview Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 41 Determining Roll Angle y Roll is the angle between the sensor y-axis and the vector from the center of the sensor frame to the center of the least squares circle, (xc, yc) x Ο æ ö -1 xc f = -tan ç ÷ è yc ø (xc, yc) 7/12/2016 Center of Least Squares Circle University of Colorado Aerospace Engineering Sciences 42 Determining Pitch Angle y Calculate Height: Height = pixelPitch (Vc - Rc ) Calculate Pitch Angle: x æ Height ö q = -tan ç ÷ è FocalLength ø -1 Vc - Rc Pitch Angle Height FOV Vc Rc Focal Length Earth center 7/12/2016 University of Colorado Aerospace Engineering Sciences 43 Algorithm Overview Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 44 Pitch and Roll Errors Pitch and roll errors are negligible after error reduction equation. DR.2.2.1 Displacements have errors less than 0.5o Error correction polynomial reduces errors to ≤ 0.06o Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 45 VERIFICATION AND VALIDATION Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 46 Verification & Validation Overview Hardware Components Hardware-Algorithm Communications Algorithm Full System Test Validation Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 47 Algorithm Algorithm Generate Perturbed Images Feed Images To Software Calculate Vector Compare with actual vector Compute Error Generate Error Model Purpose: Validate algorithm accuracy and reduce generated errors. Test for required perturbations and verify error is within 0.5o of accuracy. Requires: – Generated images of perturbations – Computer running correct software • Current error in the algorithm is less than 0.06o Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 48 Hardware Components Hardware Isolate Component Supply Power Measure Voltage Measure Wattage Purpose: Verify startup functionality and power draw of each component before and after integration. Ensure correct input voltage to sensor and microcomputer. Sensor 5V Power Regulation Board Requires: – – – – – – Outline 7/12/2016 Camera Microcomputer Microcomputer Power Regulation Board Inclinometer [collects camera angle data] Voltmeter Controllable Power Supply Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm 22-34V 5V DR.3.2.3 Power draw less than 5 W V&V Planning 49 Hardware-Algorithm Communication Isolate Component Verify Data with Computer Send and/or Check Data Further Integrate Component HardwareAlgorithm Test Data Speed and Storage Size Purpose: Ensure electrical components properly send and receive data. Verify that the output from camera and housekeeping data can be properly saved. Verify the storage capacity of the microcomputer. Requires: – Camera – Health and Safety Sensors – Microcomputer – CANBus [USB Adapter] – Inclinometer [Full Test] – External Computer Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 50 Full System Test Construct Setup Capture Data Compute Displacement Compare To Inclinometer Adjust Angle Full System Test Purpose: Final verification of ETHOS. Verify horizon tracking software in a simulated orbital environment. Ensure displacement vector is within accuracy requirements. Requires: – – – – Sensor Test Stand Earth Disk Test Enclosure Test Scenarios: – Zero angular perturbations – Pitch and roll perturbations Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 51 Camera Field of View α Altitude Boresight Horizon Line Horizon Re Disk Radius α Focal Point Height Horizon Disk Radius FOV • Camera FOV: 63o Outline 7/12/2016 Solution CPEs Camera Horizon Disk Edge Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 52 Test Stand Scaling α Boresight Focal Point Height Horizon Disk Radius Outline 7/12/2016 Altitude Scaled Horizon Radius Focal Point Height Alpha * Measured by inclinometer 250 km 16” ± 0.30” 4.52” ± 0.09” 15.79° 750 km 9.07” ± 0.11” 4.52” ± 0.09” 26.52° Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 53 Controlling Roll • Round face of sensor box is press-fit into ball bearing • Holes and pin on bearing mounting bracket fixes angle Thin-Section Ball Bearing • 5° nominal increments, actual angle measured by inclinometer Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 54 Controlling Pitch • Ball bearings allow steel rod to rotate • Rod deflection is < 0.002”, and is considered negligible • Angle is set by fixed wheel and pin Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 55 Emissivity Stefan-Boltzmann Law π π = εσπ 4 π2 • Disk coated with enamel black paint Ζ = 0.80 • Background covered in aluminum foil • Camera Sensitivity: T = 50 mK Ζ = 0.04 • Emissive power difference: Required flux to flip a bit j = 304 W/m2 j= 3.54 x10-13 W/m2 Difference between Earth and space = 360 W/m2 Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 56 Validation Angular Error Source Error Magnitude (°) Pixel 0.28 Inclinometer 0.14 Max Slew Rate 0.05 Velocity 0.03 Focal Point Height 0.29 Algorithm 0.06 Total Error 0.44 Outline 7/12/2016 Solution CPEs Camera Electrical Validation DR.2.2.1 Displacements have errors less than 0.5° Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 57 Project Risks High CAN Bus Protocol Broken Camera Error Reduction Model Lens Distortion Error Occurrence SD-Card Interface Data Processing Design/Build Power Distribution Board Power Distribution Board Failures (Power Surge or Voltage Regulator Failures) Camera Interface Find Focal Point Low Low Outline 7/12/2016 Solution CPEs Severity Camera Electrical Software University of Colorado Aerospace Engineering Sciences High Algorithm V&V Planning 58 Organizational Chart Customer Joe Breno and Randy Owen Advisor Project Lead Dr Farnsworth Mechanical Lead Manufacturing Lead Thomas Konnert Outline 7/12/2016 Solution Matt Cirbo Electrical Lead Jesse Keefer Matt Busby CPEs Systems Lead Neal Stolz Software Lead Algorithm Lead Pat Klein Taylor Dean Camera Electrical Software University of Colorado Aerospace Engineering Sciences Safety/Test Lead Financial Lead Cole Oppliger Algorithm Noah Buchanan V&V Planning 59 Work Breakdown Structure ETHOS Electrical Mechanical Algorithm Software Test Future Class Deliverables Power Distribution Wiring Diagram Gimbal System Design Horizon Detection Power reset initialization System test plan Manufacturing Readiness Review I/O Wiring Diagram Test Stand Design Pitch/Roll Determination Sensor Interface Budget/Timeline Test Readiness Review Power Distribution Board Housing Unit Design Error Model Local memory management Risk analysis Matrix AIAA Report Bit-bang design Construction Calibration Suite CANbus Protocol Implement test plan Spring Final Review Test Results Design Symposium Spring Final Report Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 60 Work Plan Financial Algorithm Test Mechanical Electrical Software Uncertainty Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 61 Work Plan WEEK 1 Essential purchases, error model added, designs finalized, sensor interface analysis, software initialization Financial Algorithm Test Mechanical Electrical Software Uncertainty Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 62 Work Plan WEEK 7 WEEK 1 Power board construction, manufacturing, sensor interface code, software interface, algorithm optimization Financial Algorithm Test Mechanical Electrical Software Uncertainty Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 63 Work Plan WEEK 7 WEEK 1 WEEK 12 Subsystem testing, full software integration, calibration suite design Financial Algorithm Test Mechanical Electrical Software Uncertainty Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 64 Work Plan WEEK 7 WEEK 1 WEEK 12 WEEK 15 Full system testing Financial Algorithm Test Mechanical Electrical Software Uncertainty Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 65 Work Plan WEEK 7 WEEK 1 WEEK 12 WEEK 15 Critical Path Financial Algorithm Test Mechanical Electrical Software Uncertainty Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 66 Cost Plan Component Price Tau2 Low Res $2,200 Housing Materials $175.00 BeagleBone Black $60.00 16 GB microSD Card $8.00 Power Distribution Board $10.00 Other Electronics $130.00 Test Stand Materials $800.00 Mylar Tent $110.00 Total $3,490.00 Outline 7/12/2016 Solution CPEs Camera • Budget Margin ≈ $1,500 • Includes estimated costs of screws, bolts, etc. • Includes shipping estimates • Backup Visual Camera = $55 – Electrical Does not satisfy IR requirement Software University of Colorado Aerospace Engineering Sciences Algorithm V&V Planning 67 Budget Sensor and Housing Components Electronics 32% 48% Testing Materials 18% Component Cost Sensor and Housing $2,400 Electronics $200 Testing Materials $900 Margin $1,500 Margin 2% 7/12/2016 University of Colorado Aerospace Engineering Sciences 68 Mass Budget 12% 12% 7% Camera Housing BeagleBone Black 69% 7/12/2016 Component Mass (g) Camera 72 Housing 412 BeagleBone 40 Margin 76 Margin University of Colorado Aerospace Engineering Sciences 69 7/12/2016 University of Colorado Aerospace Engineering Sciences 70 Backup Slides This page intentionally left blank 7/12/2016 University of Colorado Aerospace Engineering Sciences 71 References [1] http://wagner.edu/hawktalk/goddards-testing-facilitieswhere-we-build-satellites-then-try-to-break-them/ 7/12/2016 University of Colorado Aerospace Engineering Sciences 72 Risk Mitigation Risk: Occurance Severity Total Mitigation Visual Camera Backup. The IR Camera could break through 10 handling or if a power surge would occur. Camera Breaks Design/Build Power Distribution Board Cant Process data quick enough 4 5 3 4 2 2 Error Reduction Model 3 3 Power Distribution Board fails 2 4 CAN Protocol 4 3 SD-Card Interface 2 3 Other possible methods to save data. Also, sources to help if there 6 are issues. Focal Point of Camera 1 2 Camera should give a focal length, but we can find it if need be. 2 Camera distributer also can help 12 Talk to Professionals Can Negotiate on accuracy. Also now using Beagle Bone Black 4 which is very fast. Cannot detect roll accurately without this. May be negotiable. 9 Create using pictures of known angles. Preliminary model in place. 8 Safe to mate test and some parts are cheep enough to buy again Negotiable, One team member knows some on this. Others are learning about it. Professionals at LASP and Surrey know more 12 about this. Camera Interface 2 3 Camera should give the interface. We need to know if we can interface to whatever it has though. Appears as if it is possible. Off 6 ramp use camera link with frame grabber. Lens Distortion Error 3 3 This is unknown. Could be found once a camera is known. Camera 9 can correct for this. Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 73 Transformation Matrices • Roll Rotation: 1 – π π₯ π = 0 0 0 cos π sin π 0 − sin π cos π z’ x – Given the roll angle π, the above matrix gives the transform from px to px’. – p represents the output vector component along the specified axis. Ο y’ x’ – ππ₯′ = π π₯ π ππ₯ 7/12/2016 University of Colorado Aerospace Engineering Sciences Roll Rotation 74 Transformation Matrices • – π π¦ π = • z’ Pitch Rotation: cos(π) 0 − sin π 0 1 0 sin(π) 0 cos π x’ Yaw Rotation: cos Ψ – π π§ Ψ = sin Ψ 0 Vectors in the y and z directions are rotated in the same manner as the x-axis. • Final vector p is in the new coordinate frame. • Even though we are not determining a yaw vector, it is necessary for the transformation. 7/12/2016 θ Pitch Rotation −sin(Ψ) 0 cos Ψ 0 0 1 • y y’ z’ Ψ z y’ x’ Yaw Rotation University of Colorado Aerospace Engineering Sciences 75 Camera Logistical Issues • Budget Limitations – – IR camera consumes ~50% of budget Backup Solution: Visual Camera • • • Schedule Limitations – – • LinkSprite JPEG Color Camera = $55 Does not satisfy eclipse functionality requirement 2 to 4 week lead time for most cores from FLIR Order by the end of December to be ready for testing by February ITAR Restrictions – – 7/12/2016 Many compatible IR cameras subject to control by ITAR (22 CFR 121.1) ITAR/EAR or export controlled component usage prohibited University of Colorado Aerospace Engineering Sciences 76 Additional Camera Options Camera ICI 9320 C ASP-HR320 core with 48° lens Spectral Range 7-14 μm 8-14 μm Resolution 320x240 384x288 FOV 40° x 30° 48° x 34.38° <1 W ≤2.5 W 34mm x 30mm x 34mm (without lens) 40mm x 40mm x 30mm (without lens) 81g 50g Price $1,995.00 $2,829.95 Lead Time 1-8 weeks 4-6 weeks Power Dimensions Weight Drawbacks 7/12/2016 • • Analog only Unfamiliar Camera Link option, requires additional hardware • • • University of Colorado Aerospace Engineering Sciences High power consumption Unknown interface Unknown lens information 77 Backup Solution: Visual Camera LinkSprite JPEG Color Camera TTL Interface### 7/12/2016 Resolution Up to 1600x1200 Frame Rate 15 Hz FOV 102° x 86° Interface UART Power ≤120 mA Dimensions 36.6 mm x 32 mm x 27.6 mm Price $54.95 University of Colorado Aerospace Engineering Sciences 78 Pixel Error Analysis πππ₯ππ πππππ = πΉππ 1 ∗ # ππ πππ₯πππ 2 FLIR Tau2 162 LWIR Camera 7/12/2016 Horizontal Vertical Pixels 162 128 FOV 63° 50° Pixel Error 0.275° 0.276° University of Colorado Aerospace Engineering Sciences 79 Inclinometer Purpose: Verify accuracy of ETHOS calculations 7/12/2016 Parameter Value Interface SPI (0.01 – 1.5 MHz) Accuracy +0.1 o Range + 90 o Vin 3.3 V University of Colorado Aerospace Engineering Sciences 80 Sensors • Sensors required for temperature, voltage, and current. – Temperature • TMP36 – Has resolution of ± 2° C for a range of -40° to 125° C (no required resolution) – 2.7 – 5.5 VDC ADC required – Scales at 10mv/°C – Voltage • MCP3002 – 2.7-5.5 VDC – 10 bit resolution – Current • ACS712 – Measures up to 5 Amps – Calibration Needed 7/12/2016 University of Colorado Aerospace Engineering Sciences 81 Error Breakdown Testing Angular Error Source Operation Error Magnitude (°) Angular Error Source Error Magnitude (°) Pixel Pixel Inclinometer Max Slew Rate 0.05 Mount angle Velocity 5e-5 Model Earth size/distance placed Latency 5e-5 Earth Oblateness 0.06 Total: Total: Combined Total: 7/12/2016 University of Colorado Aerospace Engineering Sciences 82 Required Output Speed • With a slew rate of 3°/sec and an output of 5Hz – Error becomes 0.6° at best • Now find minimum output needed – Use .28 from all error before 0.5 = .282 +π₯ 2 π₯ = .414 3°/π ππ = .414 π¦ π¦ = 7.21 π»π§ 7/12/2016 University of Colorado Aerospace Engineering Sciences 83 Coordinate Frames Image Frame (6,6) Sensor Frame Transformation Eqs: xs = xi - 0.5w (-9,7) ys = 0.5h - yi Example: w = 30 h = 26 xs = 6 - 0.5(30) = -9 ys = 0.5(26) - 6 = 7 7/12/2016 University of Colorado Aerospace Engineering Sciences 84 Roll Errors Actual Roll Angle Roll Calculations Error 0º 0.006º 0.006º 5º 5.558º 0.558º 10º 11.105º 1.105º 15º 16.565º 1.565º 20º 22.025º 2.025º 25º 27.372º 2.372º 30º 32.600º 2.600º 35º 37.738º 2.738º 40º 42.748º 2.748º Ra = 0.9334164*Rc - 0.298595 **Accuracy increases with order of fit** Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 85 Pitch Errors Actual Roll Angle Pitch Calculations Error 0º 0.213º 0.213º 5º 5.173º 0.173º 10º 10.175º 0.175º 15º 15.146º 0.146º 20º 20.056º 0.056º 25º 24.981º 0.019º 30º 30.003º 0.003º 35º 35.045º 0.045º 40º 40.013º 0.013º Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 86 Pitch and Roll Errors • Simulated images are fed through the algorithm to determine the errors due to observed eccentricity. • A polynomial function is fit to the errors. • Measured pitch and roll are fed to the functions and corrected values are returned. Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 87 Pitch and Roll Errors Outline 7/12/2016 Actual Angle Roll Error w/o Correction 0º 0.006º 5º Pitch Error w/o Correction Pitch Error w/ Correction 0.001º 0.213º 0.006º 0.558º 0.001º 0.173º 0.023º 10º 1.105º 0.007º 0.175º 0.014º 15º 1.565º 0.020º 0.146º 0.033º 20º 2.025º 0.008º 0.056º 0.006º 25º 2.372º 0.009º 0.019º 0.019º 30º 2.600º 0.007º 0.003º 0.013º 35º 2.738º 0.025º 0.045º 0.055º 40º 2.748º 0.014º 0.013º 0.014º Solution CPEs Roll Error w/ Correction Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 88 Calibration Data • • • • • Outline 7/12/2016 Zero pitch angle Zero roll angle Pitch error reduction equation coefficients Roll error reduction equation coefficients Threshold factor Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 89 Observable Horizon 750 km 250 km Observable horizon from 250 and 750 km altitude Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 90 Focal Plane Distortion Bore sight FOV ρ Focal Plane Horizon Plane • Projection onto Focal Plane Through Angle ρ Horizon Plane Outline 7/12/2016 Solution CPEs Camera Electrical • Observed eccentricity depends on pitch angle, ρ Orientation of ellipse depends on roll angle Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 91 Linear Least Squares Fit Pros Cons • Computationally efficient • Easy pitch and roll calculations Outline 7/12/2016 Solution CPEs Camera • Location of fit is highly dependent on the amount of horizon in the image • Skews pitch and roll calculations significantly Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 92 Chord Bisector Method Pros Cons • Independent of amount of edge in the image • Based on circular geometry • Computationally efficient • Highly variable answers from chord to chord • Highly coupled with observed eccentricity of the horizon curve • Harder to calculate pitch (xc, yc) Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 93 Circular Least Squares Fit Pros Cons • Independent of amount of edge in the image • Low coupling between observed eccentricity of horizon and measurement error • Based on circular geometry Outline 7/12/2016 Solution CPEs Camera • Slightly more computationally intensive Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 94 Image Preprocessing • Many cameras perform lens distortion reduction prior to outputting image. • This corrects for barrel distortion, pincushion distortion, and mustache distortion Barrel Distortion Pincushion Distortion Mustache Distortion Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 95 Horizon Gradient • • Outline 7/12/2016 The location in the horizon gradient where the edge pixels are marked depends on the horizon gradient and the threshold value. The threshold factor can be manipulated to change the location in the horizon gradient where edge pixels are marked. Solution CPEs Camera Threshold Factor = 1 Threshold Factor = 0.25 Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 96 Algorithm Sensitivity to Dynamic Range 1 Bin Difference Horizon • Edge detection functions when there is at least a 1 bin difference between foreground and background. • Image to the right shows the edge detected with only 1 bin difference between horizon and space. Outline 7/12/2016 Solution CPEs Camera Electrical Software University of Colorado Aerospace Engineering Sciences Algorithm Testing Planning 97 Power Distribution 7/12/2016 University of Colorado Aerospace Engineering Sciences 98 Interface Timing Diagrams "Tau 2 Uncooled Cores." Tau 2 LWIR Camera Cores. N.p., n.d. Web. 02 Dec. 2014. 7/12/2016 University of Colorado Aerospace Engineering Sciences 99 Interface Timing Diagram Camera Clock 10.519 MHz Camera Line Valid Pixel Data 8 lines Pixel 1 Data Pixel 2 Data PRU Clock 7/12/2016 University of Colorado Aerospace Engineering Sciences 100 Data Size Calculation Attitude Telemetry: Each axis – 16-bit/2 byte (# of axes)*(size of each axis)*(5 Hz)*(200 min) = data size ο (2) * (32 bits) *(5 Hz)*(200*60) = 480 KB Safety Telemetry: Number of Sensors – 4 (Voltage reg input/output voltage and current) (# of sensors)*(size of each datum)*(5 Hz)*(200 min) = data size ο (4) * (8 bits) *(5 Hz)*(200*60) = 240 KB Total Data Size: Attitude Telemetry Size + Safety Telemetry = Total Amount of Data 480 kB + 240 kB = 720 kB 7/12/2016 University of Colorado Aerospace Engineering Sciences 101 Power-on Initialization • • • • Set DIO Pins Configure IR Sensor Configure Timers Enable PRUs – 2x 200 MHz 32bit microcontrollers that run concurrent to the CPU • Enable CAN communication • Enable Interrupts • Start Diagnostic Routine on CPU – Sample V/I/T at 5 Hz – Save to SD Card 7/12/2016 University of Colorado Aerospace Engineering Sciences 102 Lunar Error • πΈπ = πππ 4 • πΈπ πβπππ = Total emitted radiation of π a point object, 2 πΈ 4∗π π Total emitted radiation of π a sphere, 2 π ∗π π • Ω= π΄ π2 Solid angle, area of sphere over distance to object, sr • π = πΈπ πβπππ ∗ Ω 7/12/2016 Radiative power at distance, University of Colorado Aerospace Engineering Sciences π π2 103 Lunar Error Surface Temperature of the Moon Figure 2: Lunar Surface Temperature Use surface temperature of 400 K for worst case 7/12/2016 University of Colorado Aerospace Engineering Sciences 104 Lunar Error Emissivity of the Moon Use emissivity of 1 for worst case 7/12/2016 University of Colorado Aerospace Engineering Sciences 105 Spectral Distribution of Earth [4] 7/12/2016 University of Colorado Aerospace Engineering Sciences 106 Spectral Distribution of Sun [5] 7/12/2016 University of Colorado Aerospace Engineering Sciences 107 Earth Vs. Sun Spectrum Distribution [6] 7/12/2016 University of Colorado Aerospace Engineering Sciences 108 Solar Interference [4] 7/12/2016 University of Colorado Aerospace Engineering Sciences 109 Beam Deflection • Modeled as cantilever beam • P = 3.5 lbs • L = 10.46 inches • E = modulus of elasticity = 28000 ksi • d = diameter = 0.5 inches • I = moment of inertia = • SF = safety factor = 3 • πΏ= Deflection = 7/12/2016 ππΉ∗π∗πΏ3 48∗πΈ∗πΌ π∗π 4 64 = 0.0029 inches University of Colorado Aerospace Engineering Sciences 110 Circular Mount Brackets will allow to position focal point at rotation axes 5” Diameter 0.5” 7/12/2016 University of Colorado Aerospace Engineering Sciences 111 Pin Slop • β is angular error associated with pin slop • Resolution of Inclinometer is 0.1° • Transitional fit tolerance for RC1 close sliding fit is 0.0005” -> 0.01° error Pin Slop Hole radius from focal point axis β – Not detectable by inclinometer Outline Solution CPEs Camera Electrical Software Algorithm V&V Planning University of Colorado Aerospace Engineering Sciences 7/12/2016 112 Height Measurement Focal Point Distance Relative to Rear Distance From Front of Sensor Housing Measure distance between front of sensor box and Horizon disk Total distance of disk to focal point < 1/16th” Outline Solution CPEs Camera Electrical Software Algorithm V&V Planning University of Colorado Aerospace Engineering Sciences 7/12/2016 113 Error in Disk Radius Camera Focal Point α Focal Point Height φ Error in Radius • Disk Radius = 16.00” for 250 km = 9.07” for 750 km • φ is smaller than pixel resolution for both 250 and 750 km disks • Manufacturing tolerance governed by pixel resolution • 0.30” for 250 km • Outline 0.11” for 750 km Solution CPEs Camera Electrical Software Test Horizon Half-Disk Radius Algorithm V&V Planning University of Colorado Aerospace Engineering Sciences 7/12/2016 114 Error in Stand Height β Focal Point Height α Bore Sight • |α – β| = Angular Error • Stand Height = 4.52” • Focal point height must be known to 0.09” to keep total error below 0.5° Outline Solution CPEs Camera Electrical Chang e in Height Nominal Focal Point Height Test Horizon Half-Disk Radius Software Algorithm V&V Planning University of Colorado Aerospace Engineering Sciences 7/12/2016 115