Week Update May 20, 2005

Unofficial News:

Last week's weekly update somehow got lost, or never written, so this is a two week update. In my opinion the project is progressing well. Later next month the drilling device should be attached and tested. The project now incorporates the collision detection package SWIFT. The system doesn't seem slow. However, the ultimate test will be loading the STL file and surrounding it with the voxel shell.

Voxel Headache:

The voxel shell is the main problem with the system. This is because its shape is dynamic. The STL or TRI objects are compiled by openGl using display lists. The objects are first loaded into a temporary storage, then stored using openGl display lists. The display list stores the shape at compile time. The program does not need to store the object after the shape is given to openGl. Drawing the shape is then only one line of code.

Another issue is finding which cells of the voxel object need to be updated after a collision occurs. If the sphere contacts only a part of a voxel, should it be removed? What other issues exist?

SWIFT News:

The package was incorporated into stlViewer first and then into Prototype. Under Prototype the collision detection algorithm did not seem to slow down the system. This needs more testing, but this is a very good sign. The step of voxelizing the STL object will be left out of the system unless a noticable system lag occurs.

Status:

SWIFT Collision detection is tested and working.
TRI files are used to load the objects.
The haptic device can move the cursor on screen.
HDAPI is used to update position, force and use collision detection

Issues:

Mapping haptic coordinates to the screen.
Test collision detection with haptic cursor and TRI object.
Surround STL/TRI object with voxel shell.
Moving voxel shell, coordinate system, calibration, other related issues.
Generate correct force vector.