Saron Paz Zvika Markfeld Tomer Daniel Oleg Imanilov
An open source software & hardwareopen source Combining: Arduino Arduino Sensors Android ADK Android ADK Android
A glove that translates a predefined vocabulary of the deaf and mute sign language into text
Develop a humane, easy-to-use interface allowing hearing impaired people to communicate more freely with their physical surrounding Allow for new interaction models between (non- disabled) people and machines For example, we can all probably think of reasonable ways to express concepts such as "Send", "Undo", "Okay" etc. in hand gestures. Incorporating these into everyday interactions between people and computers may help people get into more intuitive touch with technology This, in turn, may enable elderly people and/or people coming from non-technological background to tap into today's technological world
Lily-Pad Arduino samples sensors attached to glove ADK board relays data from the Lily-Pad to Android Android uses a back propagation Neural Network to analyze data
Arduino compatible microcontroller board designed for wearables and e-textiles
Flex Sensors measure finger bending Wii Nunchuck = Accelerometer + Gyros Conductive tape detects fingers touching
Flex Sensor changes resistance when bent Using a voltage divider the voltage is read by Arduino ADC (10 bits)
2 Buttons Joystick Accelerometer (X,Y,Z) Gyros (Roll, Pitch) I2C & Power
Interconnected group of artificial neurons Good at pattern recognitionpattern recognition
Each connection has weight, which can be modified so as to model synaptic learning Each neuron computes a threshold function of the weighted sum of its inputs In a back propagation network weights are modified proportionally to the error
Artificial Neural Network GUI Relay ADK library Sensor Control Serial
How to position flex sensors? Fix tip to top and pass through sewn loops How to distinguish between R,U,V,S,E,T,N? Conductive tape detects fingers touching How to distinguish between each letter? Time based How to record movement? Freeze in the middle
The user is required to calibrate the glove once the application starts Calibration is needed to compensate for different wearers and even for re-wearing All sensors are flexed to find working minima and maxima After calibration all sensors are normalized
The user is required to train the neural network to the vocabulary Training involves recording of every gesture as well as training the neural network
Saron Paz is an experience designer & a hopeless idealist Zvika Markfeld is a design freshman and an aspiring inventor Tomer Daniel is a software architect by day & a maker by night Oleg Imanilov is a software engineer & a brilliant ideas generator
Show & Tell Google code project translator-glove/ translator-glove/ Sensors Flex Sensors Nunchuck controller-for-wii Arduino Wiichuck class Android ADK