Page 1 of 1

Patents on physics simulation / collision detection

Posted: Sun Feb 19, 2006 6:04 pm
by Erwin Coumans
Below are a couple of patents and patent applications related to physics simulation. See also the discussion on gamedev: ... _id=376063

Projected Gauss Siedel iterative lcp solver: ... a&RS=ageia
This patent pre-dates the ODE quickstep (projected Gauss Seidel) development but it describes the same algorithm and optimizations.

Russel Smith (ODE) also holds a patent on direct LCP methods (using Dantzig) ... acoursiere

This patent by John Nagle ( covers most workable methods of spring/damper (or "penalty method") simulations of jointed, colliding bodies: ... /6,067,096

Havok (Telekinesys) Rigidbody Friction Patent:

Havok Softbody patent:

A few more from
Midway ghost vehicles:
5,269,687: ... /5,269,687and 5,354,202: ... /5,354,202

Baraff, Witkin, Kass (collision detection, cloth: ... =IN/baraff

Baraff, Witkin, Kass (cloth, hair): ... =IN/baraff

Filed (RPA's: Regular Patent Applications, not yet granted):

Synthesized engine sounds for games: ... +AND+games

Physics Engine (impulse-based constraint solver): ... +AND+games

PPU hardware: Aegia: ... +AND+games

PPU hardware: Aegia: ... +AND+games

Parallel LCP solver for physics engines (Aegia): ... +AND+games

Particle System patent: ... +AND+games

Posted: Mon Feb 20, 2006 7:35 am
by Erin Catto
Just to be clear, not all of these are patents. Many are just applications.

Posted: Mon Feb 20, 2006 9:11 am
by Eternl Knight
I would also recommend that no-one read them. Regardless of the fact that patenting physics algorithms is questionable at best, reading them will allow any litigators to use "willful infringement" against you.

I will not argue the pros/cons of software patents, but every lawyer I have talked to about them have given me the same advice - NEVER read them.


Posted: Mon Feb 20, 2006 11:06 pm
by dog
Ignorance is no defence. Really. You may as well read them.

That some of these things are patented is outrageous (especially the ones I have prior art on :-) ). And we shouldn't have to be worrying about them.

If you are concerned about breaking patents, then you, as an engineer, are responsible for making sure you do your very best not to. That means knowing the patents and doing something different. Saying you didn't look has no legal standing whatsoever.

Posted: Tue Feb 21, 2006 8:31 am
by Eternl Knight
Actually, with patents - it does. If someone suing you over a patent can prove you were aware of the patent - you are liable for triple damages. This is known as "willful infringement".

Ask any intellectual property lawyer about software patents and they will advise you to do the same as I have - to not read them.

Don't believe me, check for yourself. It's no skin off my nose.


Posted: Tue Feb 21, 2006 7:56 pm
by dog
Proving "willful infringement" just requires the lawyers to state that the patents were publicly available. You don't need to see them. According to our intellectual property lawyers "ostrich-like head in the sand behavior just indicates to the judge that the engineers knew they were potentially infringing", and bingo you're hit with willful infringement anyway.

And yes it does have a chilling effect on innovation.

Posted: Wed Feb 22, 2006 1:21 am
by Eternl Knight
That makes no sense. Patents are by definition made public. That is the reason for their existence. If they're not publically available - they are not a patent, but either an application for one or a trade secret. The Patents Database is open for anyone to search - and hence is "publically available".

Talk to a lawyer about this and ask them to describe to you what constitutes "willful infringement". It is not simply the fact that the patents are publically available (as that is implicit by them BEING a patent).


Posted: Wed Feb 22, 2006 2:06 am
by dog
You may well be right. I'm just passing on advice I received from some extremely well paid attorneys.

In addition, if defending against a patent claim, you will be penalised if you can not demonstrate that you did due diligence in ensuring you were not infringing.

Basically, you can't win.

Posted: Wed Feb 22, 2006 2:33 am
by Eternl Knight
I'm not saying you can WIN based on not reading the patents, but you will be liable for only 1/3 the damages. Given the language most patent applications are written in - there is no way you can economically perform due diligence when searching for software patents (pharmaceutical & "physical" engineering patents are easier to classify). Combine that with the fact that only the US & Japan unequivoably accept software patents. Europe does not "as such" (meaning that while you may obtain the patent - enforcing it in a court of law is very much touch and go), while China and many other nations scoff at the idea.

The only way to win in a patent case if by having the patents overturned, and not even that works all the time - look at the "'Blackberry Case". The patents have been judged by the USPTO as invalid and yet RIM is still able to push for an injunction.

I'm just passing on advice from my "well paid attorneys" as well. Reading the above linked patents (and/or the applications) means you not only know about their details, but will hence be obliged to license them if you ALREADY use the methods in your products (either that, or you can pay to get the patent challenged if you can find the prior art - with no guarantee of success).