Spring Final Review. Austin Anderson Geoff Inge Ethan Long. Gavin Montgomery Mark Onorato Suresh Ratnam. Eddy Scott Tyler Shea Marcell Smalley

Transcription

1 Spring Final Review Austin Anderson Geoff Inge Ethan Long Gavin Montgomery Mark Onorato Suresh Ratnam Eddy Scott Tyler Shea Marcell Smalley Project Customer: Dr. Eric Frew Project Advisor: Dr. Ryan Starkey 8/24/2014 Aerospace Engineering Sciences - Scout Slide 1

2 Background and Purpose Autonomous search and rescue multi-copter Capable of exploring dangerous urban environments Reduce risk to human life Map the environment Navigating through doorways is a critical capability 8/24/2014 Aerospace Engineering Sciences - Scout Slide 2

3 Concept of Operations Top View Level 1 Objective: Sensing ±6cm 1m Wall Doorway Floor Measure altitude and relative position with ±3cm Level 2 Objective: Motion Side View Maintain hover ±6cm Control position with ±6cm Doorway Wall Level 3 Objective: Doorway Search and Maneuver 1m Floor Search and fly through doorway 8/24/2014 Aerospace Engineering Sciences - Scout Slide 3

4 Multi-copter CAD Design 8/24/2014 Aerospace Engineering Sciences - Scout Slide 4

5 FBD/General Design Solution 8400 mah LiPo Battery Regulates Voltage for Sensor Suite 11.1 V Battery Elimination 5.0 V Circuit Relays Power to Components and Hosts Connections Breakout Board 3.3 V LDO Regulator 3.3 V Used for Vertical Positioning Ultrasonic Sensor Power Sensor Suite 11.1 V Used to Control Mutli-copter 11.1 V Control System Multi-Copter Regulates Power for APM APM Power Module 5.3 V APM Autopilot PWM Signals to ESC RTF X8 Multi-copter Responsible for Maneuvering and Navigation USB 2.0 MAVLink Data Packets Inverter UART Ports Single Board Comp. SD Memory Card 2 Port USB HUB USB 2.0 Port TTL Level Inverted Lateral Position Data 5.0 V Processes Lateral and Vertical Positional Data and Formulates Commands Used for Lateral Positioning TOF Camera Minnowboard SBC USB 2.0 Lateral Position Data 8/24/2014 Aerospace Engineering Sciences - Scout Slide 5

6 Single Board Computer Update Upon porting TOF camera algorithms to BeagleBone Black, the two devices were found to be incompatible BeagleBone Black has Arm A8 architecture while TOF camera s Shared Object Files require x86 architecture Performed trade study of x86 SBC s, selected MinnowBoard SBC BeagleBone Black MinnowBoard Architecture Arm A8 x86 Processing Speed Power Draw 1 GHz 1 GHz 2.5 Watts (0.5A at 5V) 12.5 Watts (2.5A at 5V) Mass 38 grams 137 grams MinnowBoard 8/24/2014 Aerospace Engineering Sciences - Scout Slide 6

7 Critical Project Elements CPE 1 CPE 2 Critical Project Element Capable of ±6cm precision control Relative position measurements must have ±3cm accuracy Consequence of Failure Without precision control maneuvering will result in crashes If position measurements are inaccurate control system will be ineffective CPE 3 10 minute endurance time Failure to meet customer requirements CPE 4 CPE 5 *(Added post CDR) All components must be within the multi-copter payload capacity Blade guards must be constructed for protection during testing Takeoff capacity critical for flight/mission success Failure to meet customer requirements 8/24/2014 Aerospace Engineering Sciences - Scout Slide 7

8 Testing Overview and Results 8/24/2014 Aerospace Engineering Sciences - Scout Slide 8

9 Vicon System Functionality IR Camera Reflective Markers Vicon streams data which can be collected Vicon Global with MATLAB Reference or C++ Frame SDK Reports object position and orientation Can also get individual marker Scout Body Frame locations frames per second Sub-millimeter accuracy verifies ± Several Different 3cm measurements IR Objects cameras can track be position defined of based reflected on the light position from of markers their IR markers 0.5 Based The rotation Body on camera frame accuracy of geometry, each object position can be of defined tracker by in the space user can be found The Vicon global reference frame can also be specified by the user 8/24/2014 Aerospace Engineering Sciences - Scout Slide 9

10 Level 1 Success Level 1 Success Sensor Suite Test Relative Position (Vertical) Test Relative Position (Lateral) Test Door Detection Test Component Testing Level 1 Objective: Sensing Design a sensor suite capable of integrating with a multi-copter platform Sensor Suite shall measure relative position* of targeted objects with an error of no more ± 3cm when located 0-1 m from the targeted object *from the sensor to a specified point on the doorway 8/24/2014 Aerospace Engineering Sciences - Scout Slide 10

11 Doorway Detection Test Purpose: Test ability to differentiate doorway from wall Related Requirement: Scout shall be able to detect a doorway Test Location: Fleming Indoor Flying Lab Testing Diagram: Wall USB Doorway Field of View Testing Procedure: 1. Establish connection between camera and computer 2. Run detection algorithm 3. Verify door can be detected 4. Repeat steps with different expected poses Single board computer Time of Flight Camera 8/24/2014 Aerospace Engineering Sciences - Scout Slide 11

12 Camera Pixel Doorway Detection Testing Doorframe Hough Detection Actual Test Setup Camera Pixel Hough Transform used to find which Purpose of Average test was Error to see Standard if the system Max could pixels Error identify correspond Min a Error Deviation to doorframe doorway or not (yes or no) Test Setup As Seen By VICON Left Results Frame 4.02 cm 0.1 cm 4.28 cm 3.72 cm Once frame found 3D data of these Right oframe Doorway 2.72 always cm detected 0.3 cm when present pixels 3.37 cm was analyzed 2.26 cm o No false positives 8/24/2014 Aerospace Engineering Sciences - Scout Slide 12

13 Relative Position (Lateral) Test Purpose: Testing ability to measure relative lateral position (X & Y) and data validity with VICON Related Requirement: Scout shall sense relative position with ± 3cm accuracy Test Location: Fleming Indoor Flying Lab Testing Diagram: Wall Single Board Computer USB Doorway VICON markers Time of Flight Camera Field of View X Y Z Testing Procedure: 1. Establish connection between camera and computer 2. Run detection algorithm 3. Output relative position data (distance measurement) 4. Repeat steps with different expected poses 5. Validate with VICON data 8/24/2014 Aerospace Engineering Sciences - Scout Slide 13

14 Relative Position (Lateral) Results Scout was held stationary at a distance of approximately 1 meter from the wall, while the TOF camera and Vicon both took position data Error average equal to 1 cm Errors can be caused by doorway construction/dimension differences Data concludes ±3 cm accuracy requirement met 8/24/2014 Aerospace Engineering Sciences - Scout Slide 14

15 Relative Position (Vertical) Test Purpose: Ability to measure relative vertical position (Z) and data validity with VICON Related Requirement: Measurement of vertical position must be accurate to within to ±3cm for all capture scenarios Test Location: Fleming Indoor Flying Lab Testing Diagram: Testing Procedure: 1. Establish autopilot, computer, Single board computer Autopilot - USB, Mavlink Ultrasonic UART, 5V TTL = VICON markers Ultrasonic Sensor X Y and ultrasonic connections 2. Run measurement algorithm 3. Output relative position data (distance measurement) 4. Repeat steps with different expected poses 5. Validate with VICON data Floor Z 8/24/2014 Aerospace Engineering Sciences - Scout Slide 15

16 Relative Position (Vertical) Results Scout was held stationary at a height of approximately 1 meter while the Ultrasonic sensor and Vicon both took position data Error average equal to 0.8 cm A scalar addition was applied to the data to account for ultrasonic sensor bias Data concludes ±3 cm accuracy requirement met 8/24/2014 Aerospace Engineering Sciences - Scout Slide 16

17 Level 2 Success Level 2 Success Controlled Maneuver Test Controlled Hover Test Thrust Profile Test Level 2 Objective: Motion The control system must control the relative position of the platform to ± 6 cm of a commanded position Scout must maintain controlled hover Scout must achieve controlled dynamic motion 8/24/2014 Aerospace Engineering Sciences - Scout Slide 17

18 Mass/Power Breakdown Component Mass [g] Mounting Time of Flight Camera 140 Ultrasound 4.3 Single Board Computer 119 (Minnowboard) Blade Guards 466 Electronics 66 Total Multi-copter 1841 Battery 614 Scout Total Component Current Voltage Power Single Board Computer (Minnowboard) Argos P-100 Time of Flight Camera MB1261 Ultrasonic Sensor 2.5A 5V 12.5W 1.5A 5V 7.5W 100mA 5V 0.5W APM 2.6 Autopilot 200mA 5V 1W Multi-copter 40A (estimate) 11.1V 444W Total 44.3A 465.5W 1.52 kg Mass: 1.91 kg 44.3 A Power: 50.4 A 8/24/2014 Aerospace Engineering Sciences - Scout Slide 18

19 Thrust Characterization Test Purpose: To characterize how commands given to Scout relate to the thrust generated Related Requirement: Scout shall be capable of controlled maneuvering Test Location: Fleming Indoor Flying Lab Testing Diagram: Wooden Support Frame Load Cell Rope Testing Procedure: 1. Calibrate Load Cell 2. Start collecting data into LabView 3. Increase thrust until load cell experiences load 4. Every 5 seconds increase thrust until max thrust is achieved 5. Use the correlation between stick command and force to characterize thrust 8/24/2014 Aerospace Engineering Sciences - Scout Slide 19

20 Thrust Profiling Results Pulley Friction R/C Controller gives set throttle commands relating to constant PWM signals. Load cell gives resulting thrust for a measured PWM signal. Allows for PWM signal to be sent through Autopilot from the single board computer. Thrust Force Counter Weight Vehicle Vehicle Weight Thrust force experienced for various, constant 3 PWM = 6.89 PWM 2 T signals desired T desired T desired * Equation implemented as output for onboard processing Vehicle Weight Load Cell

21 Ultrasonic Accuracy Sensing Purpose: To ensure that while integrated with the multi-copter, the ultrasonic sensor could accurately sense to ± 3cm. Related Requirement: Sensing shall be accurate to ± 3cm Test Location: Fleming Indoor Flying Lab Testing Diagram: Propeller Wash Floor Vicon Cameras Ultrasonic Wave Testing Procedure: 1. Record ultrasonic data with propellers off 2. Compare taken data to data from Vicon 3. Get offset from compared data 4. Turn propellers on and ensure that data does not leave ± 3cm accuracy 8/24/2014 Aerospace Engineering Sciences - Scout Slide 21

22 Controlled Hover Testing Purpose: To ensure that Scout can hover at ± 6cm of desired position Related Requirement: Scout shall be controlled to ± 6cm of commanded position Test Location: Fleming Indoor Flying Lab Testing Diagram: Floor 1m ±6cm X Z Y Testing Procedure: 1. Have manual control on standby 2. Algorithm should command for 1m hover 3. Data will be logged to onboard SD card 4. Validate data with VICON system 8/24/2014 Aerospace Engineering Sciences - Scout Slide 22

23 Controlled Maneuver Testing Purpose: To ensure that Scout can be controlled to ± 6cm of desired position Related Requirement: Scout shall be controlled to ± 6cm of commanded position Test Location: Fleming Indoor Flying Lab Testing Diagram: Wall Floor >1m 0.5m 1m Z X ±6cm Y Testing Procedure: 1. Have manual control on standby 2. Algorithm should command navigation to 1m away from the wall 3. Scout shall maintain distance for 1 minute 4. Algorithm will command navigation to the center of the doorway 5. Data will be logged and compared to Vicon 8/24/2014 Aerospace Engineering Sciences - Scout Slide 23

24 Level 3 Success Level 3 Success Doorway Navigation Test Level 3 Objective: Doorway Searching & Maneuvering Search for doorway, measuring 0.9m X 2.0m, through lateral movement along wall Navigate and maneuver through a doorway upon detection 8/24/2014 Aerospace Engineering Sciences - Scout Slide 24

25 Doorway Navigation Test Purpose: Test Scout s ability to detect and navigate through a doorway Related Requirements: Scout shall be capable of detecting and navigating through a doorway from 1 meter away Testing Location: Fleming Indoor Flying Lab Procedure: 1. Full system test (with VICON trackers) 2. Ensure connections are made between major components 3. Arm the system 4. Have manual control on standby 5. Algorithm will command Scout to position itself 1m from the doorway 6. System will hold position for 1 minute to ensure sufficient data 7. Scout will then rotate 90 and navigate through doorway while capturing data 8. Data will be logged to onboard SD card 9. Validate data with VICON system Wall Floor Testing Diagram Doorway = 1m 0.5m Top View 1m ±6cm X Y Z 8/24/2014 Aerospace Engineering Sciences - Scout Slide 25

26 Systems Engineering 8/24/2014 Aerospace Engineering Sciences - Scout Slide 26

27 Systems Engineering Approach Scout Project Completion Prove Clarification Multi-copter, Doorway Performed Necessary Verification Detection by needed Blade capabilities Full on positioning TOF guard Testing customer components Breakout Scout camera board doorway sensors, and for mounting plate System search expectations single ultrasonic Propeller sensor voltage board and suite functionality distribution output and fabrication understanding navigation System computer, interference control testing system testing testing Electrical circuit Critical Controlled subsystem autopilot Autopilot Single Vicon Payload Test board project capacity trade IMU facility fabrication element hover/maneuver understanding studies sensor computer capabilities testing performance Software relative identification testing Success Major characterization communication component objectives with state Relative data defined interface Autopilot autopilot/sensor position verification PID gain collection/processin testing (data tuning verification rates, g, control algorithm communication Single board development protocols, computer data power input) processing Future project applications: Full building searches, 3D mapping, testing on various multi-copters 8/24/2014 Aerospace Engineering Sciences - Scout Slide 27

28 Systems Engineering Issues 1. Software interface problems BBB unable to communicate with Argos P100 camera Led to a late design decision to switch to Minnowboard single board computer Called for new mounting and electrical design Led to new interface challenges between the Minnowboard and other components (sensors and APM) 2. Blade Guard requirement introduced after CDR Influenced project significantly Reduced time available for other parts of project Large influence on budget, scheduling, requirements development, testing, multicopter payload capacity 3. Payload-endurance issue expected a 10 min flight duration with 1.5 kg payload. New battery research and testing Electrical design and component connection changes 8/24/2014 Aerospace Engineering Sciences - Scout Slide 28

29 Lessons Learned 1. The design never works perfectly Off-ramps are critical Design has to be flexible, more than one way to complete objectives Don t overlook the little things 2. The more testing the better More testing means problems can be detected earlier and easier 3. Clear, concise definition of design requirements Avoid confusion, time delays and inadequate design 4. Design is iterative re-design is not failure 8/24/2014 Aerospace Engineering Sciences - Scout Slide 29

30 Project Management 8/24/2014 Aerospace Engineering Sciences - Scout Slide 30

31 Management Approach Project Management Organization Project Goals 3 Levels of Success Design requirements Team member communication Master task list Planning Schedule Manpower Equal distribution of team strengths and interests Facilities Budget Monitoring Weekly Advisor meetings Weekly team updates Reallocation of manpower as necessary Comparison of planned vs. actual progress Lessons Learned Careful definition of success to ensure meaningful and achievable project results Margins are absolutely necessary Communication is key, don t assume everyone knows what you know Equal involvement from whole team No On Track? Yes Project Success Subsystem Communication Systems Approach Iterative Design 8/24/2014 Aerospace Engineering Sciences - Scout Slide 31

32 Project Budget Main Budget Breakdown 16% 20% 19% 9% 36% Mechanical Software/Controls Margin Electrical Testing Differences: addition of requirements, backup components, repairs, additional testing equipment, design changes 8/24/2014 Aerospace Engineering Sciences - Scout Slide 32

33 Industry Cost Description Amount Cost Man Hours* 5,184 hrs $162,000 Overhead 200% $324,000 Sr. Projects Budget $5000 $5,000 Multi-copter Budget $3000 $3,000 TOTAL $494,000 Calculated using average of 18 hours per week per person Cost then based on hourly wage derived from average entry level salary for B.S. in AES ($65K) 8/24/2014 Aerospace Engineering Sciences - Scout Slide 33

34 References 1 Hee Jin Sohn; Byung-Kook Kim, "A Robust Localization Algorithm for Mobile Robots with Laser Range Finders," Robotics and Automation, ICRA Proceedings of the 2005 IEEE International Conference on Robotics, pp.3545,3550, April Steux, B.; El Hamzaoui, O., "tinyslam: A SLAM algorithm in less than 200 lines C-language program," Control Automation Robotics & Vision (ICARCV), th International Conference on, pp.1975,1979, 7-10 Dec Bachrach, A.; de Winter, A.; Ruijie He; Hemann, G.; Prentice, S.; Roy, N., "RANGE - robust autonomous navigation in GPS-denied environments," Robotics and Automation (ICRA), 2010 IEEE International Conference on, pp.1096,1097, 3-7 May Laser Scanners, TiM3xx / TiM31x / Indoor / Short Range, SICK Sensor Intelligence., [Cited 10 October 2013] 5 Mid range distance sensors, Dx35 / DS35 / IO-Link, SICK Sensor Intelligence., [Cited 10 October 2013] 6 AT: Samsung Li-Ion Cylindrical 7.4V 2800mAh Flat Top Rechargeable Battery w/ PCM Protection, All-Battery.com, Total Power Solutions, [Cited 13 October 2013] 7 BeagleBone Black, beagleboard.org, [Cited 7 October 2013] 8 URG-04LX-UG01 Product Information, Hokuyo Automatic Co., [September 23, 2013] 9 MB1043 HRLV-MaxSonar -EZ4? Product, MaxBotix, [September 27, 2013] 10 3DR RTF X8, 3D Robotics UAV Technology, [cited 22 September 2013] 11 APM 2.6 Set (external compass), 3D Robotics UAV Technology, [cited 25 September 2013] 12 Laser Grid GS1, GhostStop Ghost Hunting Equipment, [cited 10 October 2013] 13 Notch Filters, Thor Labs, [cited 10 October 2013] 14 X8 Motor Out Test, YouTube.com, [cited 4 October 2013] 8/24/2014 Aerospace Engineering Sciences - Scout Slide 34

35 Appendix 8/24/2014 Aerospace Engineering Sciences - Scout Slide 35

36 Critical Design Requirements Design Requirements DR1 Sensor suite measures relative position while within 1m from the wall at an altitude of 0-2m Parent Requirement CPE 2 DR2 Scout shall be able to detect a doorway CPE 2 DR3 Relative position measurements accurate to within ±3cm CPE 2 DR4 Controlled maneuvering within ±6cm of commanded position CPE 1 DR5 All components must be within the multi-copter payload capacity CPE 4 DR6 Onboard power supply must meet 10 minute endurance time CPE 3 3/3/2014 Aerospace Engineering Sciences - Scout Slide 36

37 Ultrasonic Data [cm] Ultrasonic Data [cm] Prop-Wash Interference Results 54 Propellers Off Time [1/10 s] Propellers On Time [1/10 s] Propeller Interference Testing Shows propellers do alter data, with a mean of 52.7 cm and a standard deviation of 0.88 Will not vary the data enough for requirement failure Propeller Test Stand 8/24/2014 Aerospace Engineering Sciences - Scout Slide 37

38 Systems Summary 8/24/2014 Aerospace Engineering Sciences - Scout Slide 38

39 Success Objectives Level 1 Objective: Sensing Design a sensor suite capable of integrating with a multicopter platform Sensor Suite shall measure relative position* of targeted objects with an error of no more ± 3cm when located 0-1 m from the targeted object Level 2 Objective: Motion The control system must control the relative position of the platform to ± 6 cm of a commanded position Scout must maintain controlled hover Scout must achieve controlled dynamic motion Level 3 Objective: Doorway Searching & Maneuvering Search for doorway, measuring 0.9m X 2.0m, through lateral movement along wall Navigate and maneuver through a doorway upon detection *from the sensor to a specified point on the doorway 8/24/2014 Aerospace Engineering Sciences - Scout Slide 39

40 Payload Capacity Testing Endurance vs. Payload Attachment Test Results Attached Payload [g] Endurance Test g 12 min 31 sec Test g 12 min 27 sec Test g 12 min 25 sec Test g 12 min 30 sec Testing to ensure new battery could meet 10 min endurance requirement Lasted for 12 min 31 seconds with 474 g payload estimation Test did not include current draw from sensor suite (minimal compared to multi-copter) Further testing will be conducted 8/24/2014 Aerospace Engineering Sciences - Scout Slide 40

41 Preliminary TOF Camera Test Testing For Ability of the time of flight camera to measure wall from 1 meter away (DR 1) General Procedure 1. Time of Flight Camera placed at a distance 1m away from test wall/doorway 2. Data was captured by the camera 3. Captured data run through MATLAB script, pixels and distance measurements plotted. Results Camera is capable of producing measurements within 1m and further. Time of Flight Camera Range Testing

42 Ultrasonic Sensor Ultrasonic sensor was communicated with using a serial adapter and simulating terminal using RealTerm. Data was returned over the TX line of the ultrasonic sensor Expected ASCII letter R, followed by 3 numbers corresponding to the distance [cm] 8/24/2014 Aerospace Engineering Sciences - Scout Slide 42

43 Data Rates Scout speed = 0.2 m/s (predefined) To be within ±6cm, new position data needs to be acquired, processed and command given in: t = = 0.3s Rate = 1 = 3.33Hz (At least) 0.3 Camera updates data at 10Hz APM can be commanded UP to 100Hz With safety factor, command rate is chosen to be 5Hz

44 8/24/2014 Aerospace Engineering Sciences - Scout Slide 44