140 likes | 339 Views
Gesture Recognition System. Chuck Brazzel Vic Mata Advisor: Larry Sieh. Gesture Recognition. Gestures can be used to represent some complex action Software can be programmed to recognize certain gestures By correctly interpreting gestures, the software can signal to execute actions.
E N D
Gesture Recognition System Chuck Brazzel Vic Mata Advisor: Larry Sieh
Gesture Recognition • Gestures can be used to represent some complex action • Software can be programmed to recognize certain gestures • By correctly interpreting gestures, the software can signal to execute actions
Interfacing • Gesture Recognition System was designed to interface with IVC robot • Gestures are used to signal for robot to move on-board camera • Gestures will allow better placement of camera to improve robot navigation through obstacle courses
Requirements • To do
Implementation • Gesture Recognition System uses a combination of hardware and software • Hardware is a touch screen FTIR (Frustrated Total Internal Reflection) setup • Software interprets input gestures and utilizes the open source AME patterns library available from Arizona State
Implementation FTIR Touch screen Gesture Recognition Code (Laptop) Robot/ External Device
Hardware • Gesture system uses a touch screen input device • Touch screen uses FTIR (Frustrated Total Internal Reflection) of IR light • AVRcam is configured to detect IR light • Camera data sent across RS232 line to laptop to pass to gesture code
FTIR Touch Screen • FTIR setup uses a plexiglass sheet surrounded by IR LEDs • Some of the IR light is reflected internally within the plexiglass known as total internal reflection (TIR) • When a finger contacts the top of the plexiglass a portion of the light contacting this point becomes ‘frustrated’ • This frusterated IR light is reflected down at a 90° angle to be viewed by AVRcam
Touch Screen Circuit • Touchscreen is surrounded by 30 LEDs • LEDs use about 100mA current and 1.4V voltage drop • Resistors were calculated at about 7 ohms
AVRcam • AVRcam filters IR light and passes visible light • IR filter had to be removed and replace with visible light filter • Visible light filtering is important to reduce noise from ambient light sources (sunlight, fluorescent lights)
Serial Interface • Camera interfaces via RS232. System requires a serial to USB interface. • Code was added to get data from serial port and convert to x,y coordinates • Gesture code uses x,y coordinates for gesture recognition
Finished Touch screen • To do
Data Transfer Update images later • AVRcam included blob tracking feature • Camera can be set to track a specific color and sends a set of coordinates of where it found it • For us, IR light shows up as white so we track points of white
Data Transfer • AVRcam provides more data then needed. • Must determine which point is being tracked and pull out coordinates of the box bounding the blob • Center point of the box is calculated and this data is passed to the gesture code