Search found 324 matches

by Erin Catto
Wed Apr 15, 2009 9:34 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Box 2D, revolute joint effect mass matrix
Replies: 2
Views: 1604

Re: Box 2D, revolute joint effect mass matrix

bendaowei wrote: how -(r1.y i, r1.x j) * -(r1.y i, r1.x j)
becomes
[r1.y*r1.y -r1.x*r1.y]
[-r1.x*r1.y r1.x*r1.x] )
You need to use the outer product: v * transpose(v)

Code: Select all

[ r1.y] * [r1.y, -r1.x]
[-r1.x]
In 2D skew(r) becomes a 2D column vector (the first 2 rows of the 3rd column in the 3D skew matrix).
by Erin Catto
Wed Apr 15, 2009 9:22 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Limitations of modern realtime physics engines
Replies: 16
Views: 16957

Re: Limitations of modern realtime physics engines

I wouldn't be surprised that such a study exist somewhere... The problem being searching it You will likely find many failures. I've never heard of this being successful. I know of at least one high profile game that intended to make floating point computations deterministic but failed. In the end ...
by Erin Catto
Mon Apr 06, 2009 5:57 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Limitations of modern realtime physics engines
Replies: 16
Views: 16957

Re: Limitations of modern realtime physics engines

WhiteDragon103 wrote:I have my own optimization concepts which I am implementing, but the last thing I'd want to do is reinvent the wheel. However, in case my ideas are original I'd like to keep them to myself.
So you want us to help you to develop some ideas that you won't share with us? That's a hard pill to swallow.
by Erin Catto
Thu Apr 02, 2009 9:03 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Bullet's Voronoi Simplex Solver
Replies: 46
Views: 36052

Re: Bullet's Voronoi Simplex Solver

Yeah, it would just be a potential performance increase to search all the Voronoi regions for the tetrahedron case. There is nothing wrong with the Bullet approach. I've been using a similar approach until recently. I did try some of Casey's optimizations, but they can fall apart when computing the ...
by Erin Catto
Thu Apr 02, 2009 6:57 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Bullet's Voronoi Simplex Solver
Replies: 46
Views: 36052

Re: Bullet's Voronoi Simplex Solver

I'm now wondering if the geometric approach can obtain better accuracy if the winding order is known on the tetrahedron. For the tetrahedron we can compute the barycentric coordinates in two ways. Gino's version of Johnson's algorithm doesn't make use of the winding order and gives formulas like: la...
by Erin Catto
Tue Mar 31, 2009 11:35 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Bullet's Voronoi Simplex Solver
Replies: 46
Views: 36052

Re: Bullet's Voronoi Simplex Solver

I like the compactness of the simplex solver in Solid 3.5.6, but I noticed that there are no early returns. For a given simplex, all determinants are computed. On the other hand Bullet and the gilbert.c solvers have many early returns, meaning that many computations can be skipped. Perhaps this is a...
by Erin Catto
Mon Mar 30, 2009 4:23 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: GJK Warm Starting
Replies: 3
Views: 2541

Re: GJK Warm Starting

I've tried caching the simplex vertex indices to rebuild the simplex for warm starting. It works very well! In coherent scenarios GJK often converges with just one support point call. It also seems to produce a good guess when the cached simplex is non-optimal. Degeneracies are still possible. For e...
by Erin Catto
Fri Mar 13, 2009 12:19 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: GJK Warm Starting
Replies: 3
Views: 2541

GJK Warm Starting

I'm trying to reduce the iteration count in my GJK implementation when there is temporal coherence. The final search vector can be re-used as the initial search direction in the next call. With small frame-to-frame rotations (~10 degrees) this increases performance in my tests by 8-20%. However, the...
by Erin Catto
Fri Nov 07, 2008 3:48 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Box resting after 1 iteration
Replies: 8
Views: 3178

Re: Box resting after 1 iteration

Are you accumulating the impulses? Are you clamping the accumulated impulse instead of the immediate impulse?
by Erin Catto
Tue Sep 30, 2008 3:49 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Simulation of inextensible cloth
Replies: 9
Views: 5546

Re: Simulation of inextensible cloth

Real cloth is not inextensible, so why should the simulation be inextensible?

I would only use an inextensible model/solver if it leads to a computational advantage. In this case it appears that there is a disadvantage.
by Erin Catto
Wed Aug 20, 2008 9:18 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Critically Damped PD controller
Replies: 3
Views: 11160

Re: Critically Damped PD controller

Here are some more details (check my work). MSD: http://en.wikipedia.org/wiki/Damping m * vDot + c * v + k * x = 0 w = sqrt(k/m), z = c / (2 * sqrt(k * m)) vDot = -2 * z * w * v - w * w * x xDot = v Implicit Euler: v2 = v1 - h * (2 * z * w * v2 - w * w * x2) x2 = x1 + h * v2 Substitute second into f...
by Erin Catto
Wed Aug 20, 2008 5:41 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Critically Damped PD controller
Replies: 3
Views: 11160

Re: Critically Damped PD controller

Box2D's soft distance constraint is using impulses rather than forces, hence the 1/h term. With a frequency/damping ratio model you just set the damping ratio to 1 to get critical damping. To use frequency + damping ratio, you need to know the effective mass. Here's how I derived the relationship be...
by Erin Catto
Fri Jun 20, 2008 5:58 pm
Forum: Links, Papers, Libraries, Demos, Movies, Comparisons
Topic: PhD thesis on numerical methods for rigid bodies
Replies: 14
Views: 15339

Re: PhD thesis on numerical methods for rigid bodies

I'm interested in using the quaternion-based constraints that Claude outlines in his dissertation. Some of it is a bit confusing (he defines an attachment frame twice, etc). And some of the details are spread across the chapters, tucked away in long chains of proofs. Does anyone on this forum have e...
by Erin Catto
Sat May 31, 2008 4:52 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: A new PGS solver
Replies: 19
Views: 9847

Re: A new PGS solver

given that the task is to find a set of active constraints that will solve the problem it would seem more logical and efficient to start with the simplest cases and increase the size of the system only if a solution has not been found? this should also avoid the redundant cases you mentioned i supp...
by Erin Catto
Sat May 31, 2008 12:05 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: A new PGS solver
Replies: 19
Views: 9847

Re: A new PGS solver

I took a look at this. Failure of enumeration only seems to occur infrequently. It seems to happen when the contact force and velocity are both zero at one of the two points. Round-off error creates small negative values that break the non-negativity tests. Therefore, we should use some tolerances t...