Saturday, June 12, 2010

Multiple points of reference for tracking

Beta testing revealed that the use of one IR LED for tracking served the purpose but did not exactly provide enough reference points to simulate shapes efficiently. To solve this issue two more LEDs were added to the front of the glove (see pictures below).

Three LEDs as opposed to one lets the prototype capture left and right limits to the user's hand and a center reference point.

The motors trigger upon hitting obstacles as before, but the intensity increases the deeper the user is within an obstacle

Thursday, June 10, 2010

Prototype V3 Test

Feedback intensity and patterns changed, 3 LEDs instead of 1 used to reference points.

Degrees of freedom for tilt, Prototype V3

The following video illustrates the freedom of tilt the glove permits the user.

Friday, May 14, 2010

Video capture of random test sessions

The following video captures illustrate how beta testing took place for the prototype

Note: Lighting conditions were different in some test scenarios, Wii remotes did not fail unlike the standard USB web-cameras.





Wednesday, April 14, 2010

Prototype v2 Testing Underway

Testing on the new glove has begun. The structure of the overall glove has changed too. 22 gauge stranded copper wire is used, fastened by silicon glue. This adds more flexibility to the glove as opposed to using insulating tape to hold the wires in place.

Improvements to the code:
Applet remembers where user's hand was last so just in case the camera loses track of the glove, the virtual camera won't jump about.

Videos of tests and results will be up shortly.

Friday, February 5, 2010

Tuesday, January 19, 2010

Prototype v 2

Using web cameras to capture users' hand gestures lead to many glitches. Ambient light apart from the Infrared LEDs caused the reference point to jump around causing the virtual camera within the 3D applet to move haphazardly.

To overcome the glitch I have resorted to using two Wii remotes much like web cameras to capture users' hand gestures. A tracking algorithm was not needed to be implemented as the Wiimote comes with blob tracking.

Wiimotes talk to the PC via bluetooth radio operating on the Toshiba stack. With the help of Brian Peek's managed library for the Wiimote, I am able to get the Wiimotes to communicate with a VB.Net application that uses Microsoft's XNA to render a simple 3D world.

Code for the Arduino still remains in C++ but VB.Net is able to communicate with the microcontroller via a serial port.

A demo of the applet will be up soon.