Kineo Cam announces Fast Kineo Collision Detector KCD V2.05

TOULOUSE, France: Kineo Cam has announced the new release of KCD, a software library for collision detection.

Kineo Collision Detector KCD is the high efficiency collision detector from Kineo CAM that comes as a software library, which allows an easy integration through its C++ programming API; using 3D geometrical data, KCD compares two lists of geometrical objects and quickly returns a series of results such as interferences and distances. In digital 3D mock-up, KCD performs static or dynamic clash analysis. In robotic monitoring systems, KCD enables continuous collision avoidance.

In May 2004 Kineo CAM announced its version V1R1.5 with an improvement of performance of over 50%; consecutively several major software editors adopted KCD in their CAD/CAM or 3D mock-up products.

Last year, CYBERNETIX announced the integration of KCD into CyxPro® its software platform for graphical monitoring of remotely-operated systems. “It was important for us to rely on a collision detector which is both fast and robust. We did a careful comparison of available options and selected KCD,” explained Eric Auschitzky, Nuclear BU Director.

Today, with KCD V2.05 Kineo CAM achieves a major step of performance on speed as it is multi-threaded and fully leverages multi-core systems. Indeed, collision tests are always more than twice as fast in version 2.05 compared to version 2.04.

“Tests of performance show that KCD V2.05 performs dynamic collision detections in only 114 microseconds on large 3D models of 400 000 triangles. With such models, most of the other collision checkers would be four times slower,” says Laurent Maniscalco, Kineo CAM CEO.

Memory consumption is another important parameter that is particularly optimized in KCD V2.05. Benchmarks involving KCD and other tested collision detectors show that KDS’s memory consumption is up to six times lower. On the Apollo model used for standard benchmarking of collision detectors, KCD needs no more than 24 MB of memory.

This value can even be drastically lowered by using a KCD feature called virtual polyhedrons. In this case, memory space used by geometry falls to zero and only optimized KCD structures (collision entities) benefit from available memory.

Initialization time. Any 3D software user has experienced waiting time of variable length, always unpleasant and unproductive, when using software tools. As far as applications with interference checking are concerned, this initialization time is used to compute collision entities. One can easily understand the interest of minimizing this time as much as possible. The Apollo case requires a 0.2 second initialization time with KCD, against 5 seconds for the second best tested solution.

A thread-safe mode has been introduced among other significant improvements; KCDV2.05 can run different tests over the same scene in simultaneous threads. This offers a whole new world of possibilities to multithreaded applications aimed at performance and reactivity.

For the developers, KCD V2.05 offers a fully modular collision detection and distance measurement framework. It is possible to implement proprietary collision detection primitives as new elementary computation bricks. This allows developers to just focus on the mathematical definition of the collision, while trusting KCD to manage the optimized geometry filtering algorithm. Possible applications include: molecules, laser or particle beams, non-triangulated procedural geometry, voxels, etc.

“Of course this new version of KCD is now available for the great benefit of the thousands of current end-users; and with such unmatched performances we are confident on having this number growing quickly,” says Laurent Maniscalco.