WiDraw: Enabling Hands- free Drawing in the Air on Commodity WiFi Devices
Outline Introduction System Challenges and Solutions Evaluation Conclusion
Introduction There are many ways to do hand motion tracking Is there a way to track the hand motion that is much simpler ?
Outline Introduction System Challenges and Solutions Evaluation Conclusion
Motion Tracking Ideas Use existed applications to do hand writing recognition Use existed work to do 1D AoA recognition
System Architecture
Hardware prototype on HP Envy laptops, using Atheros AR9590 chipsets and 3 antennas
Outline Introduction System Challenges and Solutions Evaluation Conclusion
Calculate the AoA
Calculate the AoA (2)
Discover and Calculate New AoA There might be new AoA for the system to track The new AoA can be calculated by the coordinate of the hands. The system needs at least two spots to calculate a spot passed by the signal beam. Using the equations below to calculate the azimuth and elevation to calculate the AoA
AoA requirements The signal strength for specific AoA couldn’t change rapidly, ex moving cell-phone, Wi-Fi AP with strength changing periodicity. How to remove it Using previous work Bringing Mobility-Awareness to WLANs using PHY Layer Information
Tracking the Hand’s Coordinate -- Tracking the Hand’s Depth Depth – the distance from hand to the line of Antenna array
What if depth change We need to know the initial depth and the time which has maximum drop at each AoA - t0, t1, t2 a = (t1-t0)/(t2-t0)
The Hand Coordinate We could calculate the distance between the hand and the trajectory of the incoming signal Using RSSI can help us know if the hand is moving away from an incoming signal or toward the incoming signal Use the above observation to calculate the hand position
An Example to track hand position
Dealing with Environmental Issues
Another way is to ignore the change if the duration between the last change is too long. T is 1.5 sec, this may lower the accuracy and yield a false positive rate less than 0.04 events/minute
Outline Introduction System Challenges and Solutions Evaluation Conclusion
Evaluation – 2D Drawing Evaluation
Evaluation – 3D Drawing Evaluation
Evaluation – Real Environment
Letter Recognition Success recognition rate for individual letter. Left number for upper-case, Right number for lowwer-case Using the writing reorganization application - MyScript Stylus
Letter Recognition Horizontal Distance: Distance from hand position to the line orthogonal with antenna arrays.
Letter Recognition Width: the letter’s width
Letter Recognition
Word Recognition According to letter reorganization result, the horizontal range is 100cm and letter width 30cm => write at most three letters at a time
Word Recognition Accuracy rate against word length
Word Recognition Accuracy Rate against different users
Outline Introduction System Challenges and Solutions Evaluation Conclusion
Conclusion
Limitation Wi-Draw using multiple antennas to track hand’s position, but most laptop has only one or two antennas. Need robust AoA strength for tracking positions