We are currently working working on a technical mixed-reality simulation which is implemented as a MDL-only level and includes following models / actors:
1. simple, static objects that represent a workpiece
2. a movable medium-poly object (30.000 polys) that represents a tool
The tool is now moved by 6 DOF data (each 3 for position and 3 for angles) that comes from a magnetic tracking system.
The same setup as in the 3DGS application is also existing as a real-life model, made out of plastics and with the magnetic tracking system attached to it.
So this real-life model in theory provides a perfect collision system alss for the 3DGS application in avoiding the tool penetrating the workpiece.
In reality the tracking system is not accurate enough to avoid any penetration.
So we had the idea to add in parallel another collision detection in 3DGS.
The requirements would be:
1. Work in conjunction with the 6 DOF data input
2. Return precise Collision at polygon-level - also for the moving object
3. Return the number and time of contacts between tool and workpiece (for generating a training score)
So far I have experimented with c_move (no hull at poly_level possible - no precise collision possible) and 3DGS-physics (strange results - how to use tracking-data in parallel to physics?).
Has anybody any advice how we could solve our problem? Saw the impressive demo of the liteC integration of the Newton engine ...
BTW: we are currently using 3DGS 6.6. Update to v7 would be no problem it it had any advantage.
