Presentation is loading. Please wait.

Presentation is loading. Please wait.

Sean Day Diante Reid Liem Huynh. Project Overview  To create a vehicle that autonomously follows a moving object using color detection  To design a.

Similar presentations


Presentation on theme: "Sean Day Diante Reid Liem Huynh. Project Overview  To create a vehicle that autonomously follows a moving object using color detection  To design a."— Presentation transcript:

1 Sean Day Diante Reid Liem Huynh

2 Project Overview  To create a vehicle that autonomously follows a moving object using color detection  To design a low cost, mobile robot that can track objects based on image processing  Implement all of the parts using the Atmel microcontroller  Track any color specified by the user

3 Requirements  Operate solely on 7.2 V Ni-Cad battery for at least 1 hour  Keep a minimum of 10 inches away from target at all time.  Have a dimension of no more than 12x15 inches  Track object within color range “0 80 0 80 50 255 “

4 Top Level Diagram CMUCam2+ Maxbotix LV-EZ2 Sensors Manager Controller Actuators Environment Images Ultrasonic Signals PWM Centroid and Servo Location Target Location Target Range PWM Software Hardware Environment Target

5 Microcontroller - Arduino  ATMEGA328  USB Interface  Cross-platform  Easy to program  Open source  Well documented

6 Microcontroller Specs  Operating Voltage : 5V  Digital I/O : 8 Pins  PWM: 6 Pins  Analog Input : 6 Pins  Flash Memory : 32kB  Clock Speed : 16 MHz  Communication : UART

7 Printed Circuit Board  PCB123 software  $100 student credit from sunstone  Prototyped on the Arduino board  2 layer design  Using through hole and surface mount techniques

8 Image Processing Goals  Color detection  Ability to distinguish between specified color and other colors in environment  Detect centroid of specified object

9 Blob Detection  Middle Mass  Determines if a group of connecting pixels are related to each other by surroundings  Efficient in identifying separate objects in a scene

10 Image Processing Techniques  Edge Detection  Canny detection  Edges are areas where a jump in intensity from one pixel to the next occurs  Able to reduce the amount of data processed by filtering out useless information

11 CMOS vs. CCD Sensor CMOSCCD (charge coupled device)  Transistor based  Flexible design  Average picture quality  Low power consumption  Low Price  Analog device  Rigid design  Excellent picture quality  Power hungry  Very Expensive

12 Choosing a Vision System  CMUcam1  CMUcam2  CMUcam3  AVRcam  Logitech QuickCam Orbit AF Webcam + RoboRealm

13 CMUcam Comparison PriceFrame RateResolutionRAMROMSPEED CMUcam1$109.9917 fps80x143136 bytes2048 words75 MHz CMUcam2$179.9950 fps176 x 255263 bytes4096 words75 MHz CMUcam3$239.9926 fps352x28864 KB128 KB60MHz AVRcam$99.0030 fps88x144700 bytes512bytes16MHZ

14 CMUcam2+ Vision Sensor Performs image processing duties for AVR Track user defined color blobs at up to 50 Frames Per Second Track motion using frame differencing at 26 Frames Per Second Gather a 28 bin histogram of each color channel Process Horizontally Edge Filtered Images

15 Why CMUcam2+  Compact Size  Frame Buffering  Affordable price  Multiple Servo Control  User Support  Histograms

16 CMUcam2+ Software  Open Source Programmable  CMUcamGUI  Allows user to see from camera’s perspective

17 Power Needs PartWatts Ultrasonic Sensor50mW H-Bridge10W Steering Servo75mW CMU cam2+1.5W

18 Voltage Regulation  All parts on AVR can run off of 5volts DC  Stepping Down 7.2 Volt battery  LM7805 regulator

19 Ultrasonic Sensor Objectives  Purpose is to keep AVR within 10 inches of target object  Be able to fit on front bumper  Will not loose the target object  Low power consumption

20 Ultrasonic Sensor Comparison Sensor ModelPriceSizeOutputPower SRF08$64.001.7” x.79” x.67” Analog/Digital75mW Ping Range Finder $29.950.85 “ x 1.8 “ x 0.6 “ PWM100mW Maxbotics$27.95.79 “ x.87 “PWM/Analog10mW

21 Maxbotics Ultrasonic Sensor  Maxbotics EZ2  Will easily fit on bumper  Only draws 2mA of current  Easy to interface

22 Interfacing the Sensor  Pulse Width Modulation  147 microseconds/inch  Analog Input  (Vcc/512)/inch  Communication  RS-232 Serial  Predetermined distance controlled by C progam

23 H-Bridge  Pololu High-Power Motor Driver 18v15  70A stall current  1.3” x 0.8”  15A continuous output  Max PWM frequency 40Hz

24 Batteries

25 Chassis  4 wheel drive  2 front turning wheels  2 rear wheels for going backward and forward  1 DC motor  1 Hobby servo  Rooms for expansion

26 DC Motor  Speed  19,200 Rpm  Stall Current  32A  No Load current  1.3A  Operating Voltage  3.6 to 6V

27 HiTec HS-311 Servo  Pan, Tilt, Steering  Max speed .15sec/60 °  Stall Torque  42 -49 oz/in  Operating Voltage  4.8-6.0V

28 Sensors Manager  getRange()  Returns range from Ultrasonic Sensor in inches  getCentroid()  Returns centroid location of target in x and y format  getPan()  Returns location of panning servo  getTilt()  Returns location of tilting servo

29 Centroid tracking  Tracking the object’s centroid 14488

30 Coordinate Transformation  CMUcam to Body alignment  Body Frame, CMUcam Frame β Servo Positions 128, 0° 210, 90°46, -90° β Servo PosOffset -90 ° 46-44 0°0° 1280 90 ° 21044 Offset = 44 sin( β ) Centroid_B = Centroid_C + Offset Negative Stop Positive Stop

31 Navigation Flowchart Initialization Forward Straight Forward Right yes no Object' s range <=10 inches? Centroid > 54 Forward Left Centroid < 34 else Stop Object' s range = 10 inches? yes Reverse no yes Object Found? noyes Good Detection? no Enter Search Mode

32 Search Mode  If no detection is found, AVR will go into search mode  Pan from negative stop to positive stop (180 ° ) Pan left Object Found? Servo Pos > -90° & < 0? no yes no Pan right Object Found? Servo Pos < 90°? no yes no Exit yes No detection

33 Fault-tolerance  AVR stores the last 10 detections into an array: 1 for hit 0 for miss  Confident level is defined by taking the average of the stored values  If confident level > 70% then it’s a good detection 1 1 1 1 1 1 0 0 0 0 1 1 1 1 0 0 1 1 1 1 Confident level= 70%  Good Detection!!

34 Proportional Controller P (Range) Plant + - InputOutput Error Variable speed depends on range from target P gains need to be tuned All control process is done through software

35 Testing  DC Motor/ H-bridge test  Range Finding Test  CMUcam2+ Pan and Tilt Test

36 Testing  Locomotion test with ultrasonic sensor stationary target  Locomotion test with ultrasonic sensor moving target  Locomotion test with CMUcam2+ with stationary target  Locomotion test with CMUcam2+ with moving target  Locomotion test with all sensors

37 Finances Part Number Part NameQty Expected Cost Real Cost Development Cost Order DateManufacturer N/A Basic Breadboard 1 $11.95 6/30/09SparkFun LV-EZ2 Ultrasonic Range Finder Maxbotix LV-EZ2 1 $27.95 6/30/09Maxbotix N/A Arduino Starter Kit 1 $49.95 6/30/09Arduino SN754410 H-Bridge 3 $12.00 --- $12.00 7/21/09Texas Instruments R245- CMUCAM2 +-Plus CMUcam2+ 1 $169.00 7/21/09Acroname FT232R USB to Serial kit 2 $29.00 $58.00 7/21/09FTDI Chip n/aBattery/charger2$55.45 8/28/09Tenergy HS-311 RC servo2$25 9/3/09HiTec ROB-09107H-Bridge1$39.95 11/01/09Pololu n/aChassis1$109.00 10/01/09Exceed-RC LM 7805Voltage Regulator6$10.00 9/01/09 Fairchild Semiconductors n/aPCB2$100.00$0.00 $100.00 11/12/09Sunstone n/aMiscellaneous17$25.00 --- Total Cost ---40 $664.25 $552.25$693.25---

38  Thank you, feel free to ask any questions…please come next door for demo


Download ppt "Sean Day Diante Reid Liem Huynh. Project Overview  To create a vehicle that autonomously follows a moving object using color detection  To design a."

Similar presentations


Ads by Google