Search found 117 matches

by ngbinh
Wed Jan 19, 2011 3:33 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Lemke's algorithm and pivoting
Replies: 7
Views: 5551

Re: Lemke's algorithm and pivoting

It seems like you want to use PATH: http://pages.cs.wisc.edu/~ferris/path/

It should be much more robust and faster than Lemke.
by ngbinh
Wed Jan 12, 2011 3:50 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Lemke's algorithm and pivoting
Replies: 7
Views: 5551

Re: Lemke's algorithm and pivoting

I meant the implementation of the work on the paper.
You can find it here: http://www.openrtp.jp/openhrp3/en/index.html

Last time I checked it contains C++ implementation of their Lemke. But keep it mind that the method is quite expensive and not suitable for game physics.
by ngbinh
Thu Jan 06, 2011 4:48 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Asking for Expertise's Opinion for "the ultimate" solution
Replies: 9
Views: 6994

Re: Asking for Expertise's Opinion for "the ultimate" soluti

I thought setting the body to be Kinematic and control it by setting velocity should work as you expect. If not, it should be simple to fix too.
by ngbinh
Thu Jan 06, 2011 4:46 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Lemke's algorithm and pivoting
Replies: 7
Views: 5551

Re: Lemke's algorithm and pivoting

I think this paper http://www.roboticsproceedings.org/rss04/p12.html could answer most of your questions. Also, you can find the implementation online ( google for it ).
by ngbinh
Wed Oct 13, 2010 9:56 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: How to solve MLCP with Projected Gauss-Seidel?
Replies: 18
Views: 10310

Re: How to solve MLCP with Projected Gauss-Seidel?

Let me put it another way: Bullet does not solve for friction and normal forces together but first assume zero (or known) friction, solve for normal forces, then plug that in to solve for friction forces. In a way, it's exactly the same as method described in http://www.cs.ubc.ca/labs/sensorimotor/p...
by ngbinh
Tue Oct 12, 2010 8:48 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: How to solve MLCP with Projected Gauss-Seidel?
Replies: 18
Views: 10310

Re: How to solve MLCP with Projected Gauss-Seidel?

I will try to answer the question: "Why there are zeroes in the diagonal?" The answer is that:...they are always there. That's the only way we can model (linearized) Coulomb friction. The third line in the matrix formulation is: -transpose(e)*f_b + mu*f_c \perp lambda, where f_b, f_c,lambda are norm...
by ngbinh
Tue Oct 12, 2010 8:21 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: How to solve MLCP with Projected Gauss-Seidel?
Replies: 18
Views: 10310

Re: How to solve MLCP with Projected Gauss-Seidel?

Thank you Dirk for your answer. http://img524.imageshack.us/img524/7986/85569142.png - Ros. This one looks like what we called Stewart-Trinkle or Anitescu-Potra time stepping method. The formulation dates back to 1996 ( Stewart-Trinke paper). Note that it's not the same formulation that Bullet, ODE...
by ngbinh
Mon Sep 20, 2010 6:01 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Movie introduction of dVC3d (using Bullet interface).
Replies: 2
Views: 2799

Movie introduction of dVC3d (using Bullet interface).

I've been working on dVC3d ( http://cgi2.cs.rpi.edu/~nguyeb2/doku.php?id=dvc3d ) for a year or so. The movie: http://www.youtube.com/watch?v=Yh5qKwK8FcQ ( It's best to view in 1080p , especially if you want to see what happen inside Blender) Short introduction: dVC3d is intended for scientific/robot...
by ngbinh
Mon Aug 16, 2010 2:45 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Chin's symplectic 4th order methods
Replies: 3
Views: 2506

Re: Chin's symplectic 4th order methods

We've been using our simulation on various robotics experiments and get close to experimental accuracy with just semi implicit Euler integrator. You may have problem with linear integrator when dealing with highly articulated bodies with many bodies connected through joints. But for that situation, ...
by ngbinh
Sat Jun 05, 2010 7:18 pm
Forum: General Bullet Physics Support and Feedback
Topic: Bullet vs ODE for simulating robotic systems
Replies: 21
Views: 15442

Re: Bullet vs ODE for simulating robotic systems

Featherstone will surely help articulated bodies. But Dantzig is unnecessary. It has been proved that with Bullet's friction model, the underlying problem is a convex QP. And convex QP solvers are much more mature than LCP ones. Also, if carefully implemented, the problem can be solved in linear tim...
by ngbinh
Sun May 02, 2010 6:32 am
Forum: General Bullet Physics Support and Feedback
Topic: 16 byte alignment issue in vector< btVector3 >
Replies: 6
Views: 9771

Re: 16 byte alignment issue in vector< btVector3 >

I am having a similar issue. I am using boost::python to wrap c++ classes which use Bullet. If any class contains a btVector3 member, it will not compile, or if I try to expose btVector3 as a boost::python object. Here's the result: C:\boost_1_38_0\boost/python/converter/as_to_python_function.hpp(2...
by ngbinh
Wed Nov 18, 2009 1:43 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Point to point constraint jacobian row count
Replies: 7
Views: 3254

Re: Point to point constraint jacobian row count

This has nothing to do with 2D or 3D. A point to point and a distance constraint are two different things. Distance: C = |x2 - x1| - L0 = 0 Cheers, -Dirk Hmm, this constraint is not linear in term of x1,x2 unless you use current values of them. But then it wont be in the Jacobian anymore as it does...
by ngbinh
Tue Nov 17, 2009 4:50 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Point to point constraint jacobian row count
Replies: 7
Views: 3254

Re: Point to point constraint jacobian row count

I'm trying to get clear on how to formulate constraints and by contrasting the Bullet btPoint2PointConstraint and Box2D b2DistanceJoint I think I am seeing two different approaches to the same problem. btPoint2PointConstraint makes use of three orthogonal (world) axes to construct 3 Jacobian rows: ...