ORBUAVC

OBJECT RECOGNITION BASED UAV CONTROL

I recieved an AR Parrot Drone as a gift during sophmore year. This was a couple months before "drone mania" spread across the tech world and media. Soon, I began hearing and reading about crazy, cool drones people were designing and programming. I wanted to join in on the fun and with my new programming skills (from AP Computer Science), I decided to program my drone to do something unseen in this rapidly advancing industry. As I read about companies' drones, I discovered one common problem: all followed people and objects by GPS rather than computer vision. While computer vision was being used in algorithms such as optical flow, I found that it was mostly being used for obstacle avoidance, rather than following or tracking. Use of object recognition would greatly enhance the drone's ability to follow people and movements precisely as well as provide a wider range of possible functionalities. Therefore, I decided to program a drone that could follow people based on recognition of an object, such as an article of clothing or a patterned patch. Depending on the color of the clothing or pattern on the patch, the functionality of the drone could change. Thus, my "Object Recognition Based UAV Control Project" was born.

VIDEOS

Object Recognition Based UAV Control (ORBUAVC) Playlist

• Object Recognition Based UAV Control (Project Summary/Demo Video as of 3/24/15)
• ORBUAVC Night Patch Detection Monitor Display 3/20/15
• ORBUAVC Night Time Patch Detection 3/20/15
• ORBUAVC Patch Detection Monitor Display 3/20/15
• ORBUAVC Patch Detection Test 3/20/15
• ORBUAVC Vision Tracking and Following 2/15/15
• ORBUAVC Flying and Tracking Test 2/13/15
• ORBUAVC Tracking & Flying (Stabilization in All Directions) 1/3/15
• ORBUAVC First Flying Test Using Computer Control 1/3/15
• ORBUAVC First Outdoor Camera Test 12/30/14
• ORBUAVC New Bounding Box and Tracking Test 12/29/14
• ORBUAVC Bounding Box Test 12/29/14
• ORBUAVC Color Thresholding Module Demo 12/7/14

PROGRESS

I have already submitted my project to the 2015 Greater San Diego Science Fair and did exceptionally well, qualifying for Intel ISEF 2015 and California State Science Fair 2015, winning Sweepstakes and 1st place at GSDSEF, and recieving several corporate awards. I am continuing research as well as preparing for ISEF, Google Science Fair, and CSSF. My drone is functional and able to track and follow a patterned patch. It can also follow a purple headband, but I have found that this color thresholding mechanism is much more sensitive to light and background noise. In my continued research, I hope to make the pattern detection algorithm more robust, implement turning (by taking advantage of the fact that the pattern is used for camera calibration), and controling the drone through EEG mind control.

AWARDS

Greater San Diego Science Fair (GSDSEF) 2015:

• Fair Awards
  • Intel ISEF Trip Sweepstakes
    • qualify for the 2015 Intel ISEF in Pittsburgh
    • one of the top six projects at the GSDSEF
    • one of the top three physical science projects at the GSDSEF
    • best computer science project at the GSDSEF
    • $2000 cash prize
  • Intel Excellence in Computer Science Award
    • best computer science project at the fair
    • Intel ISEF Affiliate
    • $200 cash prize
  • GSDSEF Senior Division 1st Place Award
    • qualify for the 2015 California State Science Fair (CSSF) at USC
    • one of the best projects among the Senior Division
    • one of the best Senior Division computer science projects
• Corporate and Professional Society Awards
  • Armed Forces Communications and Electronics Association
  • Experimental Aircraft Association, Chapter 14, San Diego
  • Marine Technology Society (ProfSoc)
    • get to present project at corporate dinner
    • cash prize
  • NAVY ONR (US Navy Science and Technology)
    • ISEF Affiliate
    • medal and $50 gift card
• Media
  • Photos:
    • GSDSEF Awards Ceremony and Media Day (3/25-26/15)
  • Links:
    • Union Tribune: Science Fair Crowns Contest Winners (3/27/15)