Open-sourcing the MuJoCo physics simulator for robotics

https://deepmind.com/blog/announcements/mujoco

by stokaasto • 5 years ago

362 69 5 years ago

jefft255• 5 years ago

This is really good news. It was a really contentious issue in RL and robotics research to be so reliant on proprietary (and very expensive) software.

asah• 5 years ago

+1 - plus, it's an immediate barrier for attracting new people to the field.

minimaxir• 5 years ago

For more context on "very expensive," a MuJoCo license was several hundreds of dollars.

cyber_kinetist4• 5 years ago

For undergrad students trying to dabble in reinforcement learning, several hundreds of dollars can be quite a lot.

GistNoesis• 5 years ago

For reference here is the price before as seen on november 2020 (previous internet archive screenshot), https://web.archive.org/web/20201111235930/https://www.robot...

TLDR: personal non commercial 500$/year for usage on up to 3 computers personal commercial 2000$/year

currymj2• 5 years ago

the price was reasonable for a good piece of software, but for the way people want to do ML research now (launch tons of parallel jobs on lots of cloud machines) the price, and more importantly, the burden of managing licenses, was prohibitive.

klowrey• 5 years ago

There was a free license for individual users, with the non-free license applicable to those who were receiving financial support (i.e. academic / industry researchers, etc.). The individual license was very popular and heavily utilized.

KaoruAoiShiho• 5 years ago

Is this a competitor to physX?

6gvONxR4sf7o2• 5 years ago

Nope. PhysX is all about realtime and gaming is the end use it was designed around. TFA has a section about why that’s bad. The different end uses mean there are miles of differences between them even though they’re both physics simulators.

sailingparrot2• 5 years ago

A lot of RL problems can be adequately simulated in either Mujoco or PhysX. So even though as you pointed out they have key differences, there is a pretty big overlap in use cases. So in that sense, I believe it still does qualify as competition.

atty• 5 years ago

I tend to be more than a little skeptical of organizations like DeepMind and (especially) OpenAI lately, when it comes to their pledges to democratize their advances in AI/ML/robotics. This, however, is 100% a great move by DeepMind in the support of the community as a whole. Kudos to the team, and a big thanks as well.

querez2• 5 years ago

Pretty sure this was just a reaction to losing so much mindshare to Brax since that came out.

Permit• 5 years ago

My understanding is that DeepMind did not previously own MuJoCo, but today they have acquired it and released it for free.

On the other hand, Brax is developed by Google and is written using Jax (which DeepMind loves and uses extensively).

Why would DeepMind have been concerned that people were switching to Brax?

nmca• 5 years ago

Brax is dope, but it's a long long way from feature parity with MuJoCo, which is world class.

modeless• 5 years ago

Great news! It was a real issue that so much RL research relied on non-free software. It was a problem that a big pile of money could solve, so I'm glad someone with a big pile of money did :)

bionhoward• 5 years ago

This is huge for the open source ML community because so many academic and commercial projects rely on MuJoCo which is out of reach for folks working on hobby projects or at tiny startups. This makes a lot more AI environments work for everyone. Major props to DeepMind

cepher• 5 years ago

I wonder what this means for the future of Brax [1].

1. https://github.com/google/brax

cyber_kinetist4• 5 years ago

Google Research already maintains the Bullet Engine as well (although it mostly seems to be Erwin Coumans maintaing it)

Each have their own strengths, there is no "definitive" physics engine. Bullet is probably the most generalist engine, but it suffers a bit because of this approach (it's split between the old 2.0 C++ API vs. the new PyBullet API with vastly different functionalities)

CamperBob26• 5 years ago

One strong advantage of MuJoCo is its plain vanilla C bindings. Hardwiring an API like this to C++ or Python makes life much harder than necessary for users of other languages. Conversely, it's very easy to wrap a well-thought-out C API in your language of choice.

Super impressed with the data structures (and documentation) they've already released.

notesonbots• 5 years ago

sfdsdf

AlexCoventry• 5 years ago

You should favorite it, if you want to leave a placeholder on it. It's a link near the top of the thread.

a-dub• 5 years ago

interesting. will they be including any biomechanical models as part of the open sourced stuff?

alhirzel• 5 years ago

Does anyone know how this compares to e.g. Chrono?

https://github.com/projectchrono/chrono

rg1112• 5 years ago

MuJoCo is the standard in much of robotics, and recently in the Deep RL space.

When I was learning Deep RL, I edged on what could be done through OpenAI gym, or self-made environments, I decided to begin working with MuJoCo.

But for an undergrad student from a developing country, the fees were extremely high.

So, I just didn't go forward with Deep RL much. I dabbled for a while and tried using PyBullet. Did not go that well.

Also thought about creating custom environments in Unity and then apply Deep RL in there. But it took too much effort.

Creating a complex enough environment and getting used to it were, in themselves- full projects that took long term commitment.

As a result, I did not learn much Deep RL to be able to apply it to complex, real-world-like scenarios.

I think that it is going to change.

I am excited about books, blogs, and videos that are going to be created as MuJoCo is open-sourced and available for free.

I, personally, will get back into Deep RL the next opportunity I get.

5faulker• 5 years ago

It would still be a while that we achieve emergence (if that's even possible) anyway, so time is kind of on our side.

EvgeniyZh• 5 years ago

Pretty sure it had free educational license tho

rg1112• 5 years ago

My uni was not rich enough to have an email server for students.

So, I did not have a uni mail address.

That was required.

CryingSofa• 5 years ago

Somewhat. The free educational license was only acceptable for 'personal' projects, so not suitable for class work.

currymj2• 5 years ago

you could install that license on one physical machine. so say you want to use a Colab or cloud GPU or something, or your university compute cluster, it wouldn’t work.

TaylorAlexander• 5 years ago

I similarly did not get in to using MuJoCo because I want all my work to be reproducible with open source. So I spent time trying to get RL working in Godot video game engine, but ran in to some road blocks. I know MuJoCo already has all the necessary pipelines so this is exciting indeed!

pol4ko• 5 years ago

Hi. Do you mind sharing more of your work with Godot? I'm working in a similar project for software testing.

siekmanj• 5 years ago

This is an amazing outcome. The license cost was a huge barrier to entry and the progress of deep RL generally.

amelius2• 5 years ago

Can it do fluid simulations for aquatic robots?

CamperBob26• 5 years ago

The docs talk about that a little [1], with the answer boiling down to "Yes, the necessary hooks are there, but no, you don't want to use this particular engine for fluid dynamics."

The docs are definitely a strong point, nicely written and presented.

1: https://mujoco.readthedocs.io/en/latest/computation.html# , about 1/4 of the way down the page

zokier• 5 years ago

It seems great, but the simulation geometry options feel bit limited? What would be really great would be the ability to simulate scenes with objects imported from STEP/IGES files, so you can do more of the modeling in actual CAD tools. Instead, MuJoCo seems more oriented towards using its own primitives, not sure how that works out with more complex situations.

ali_m• 5 years ago

If you can turn it into a collection of STL or MSH files then you can simulate it in MuJoCo. There are some caveats - MuJoCo will use the convex hull of the mesh for collisions, so if you want to accurately simulate collisions with concave features then you'll need to split your mesh into convex submeshes (there are various third party tools that can automate this). Sometimes you need to tweak the mesh geometry a bit in order to get the right contact behaviour, and in some cases it's easier to use primitives for key collision geometry and just use the meshes for visualisation.

CamperBob26• 5 years ago

They're not in the file-reading business. There's a staggering variety of 3D object file formats, both open and proprietary, so it's logical enough to let the application worry about reading the geometric data and converting it to the representation needed for physics simulation.

whinvik• 5 years ago

A little off topic but is there a good library for physics simulations that involve cloth modeling.

gautamcgoel• 5 years ago

I wish someone would do this for MATLAB.

HarHarVeryFunny• 5 years ago

I assume you know about GNU Octave? I used it years ago for Andrew Ng's ML course. It seems very good.

cyber_kinetist4• 5 years ago

Since it has a C API, I guess binding it with Matlab won’t be that hard?

amelius2• 5 years ago

MuJoCo is a second-order continuous-time simulator, implementing the full Equations of Motion

Are there any books describing how to implement this kind of physics engine?

wisty• 5 years ago

https://homes.cs.washington.edu/~todorov/papers/TodorovICRA1...

cyber_kinetist4• 5 years ago

The Mujoco documentation explains the mathematics behind it in detail.

Animats• 5 years ago

The demo object that turns itself over is nice.

I used to use a rattleback as a test back when I was working on accurate but slow physics engines in the 1990s. That's an ellipsoid cut by a plane that isn't axis-aligned. You spin it, and it spins, stops, rocks, and spins a bit in the other direction. Very similar to what they're doing.

The key to this is a multi-point contact model. If your contacts are single points, the frictional effects are wrong. It's not that hard to do this. I held a patent on that (#6,067,096), and there was a product called Falling Bodies, the first ragdoll system that really worked.

It was just too early in the 1990s. 100 MIPS was not enough.

CamperBob26• 5 years ago

It's hard to do it if some asshat patents it, that's for sure. At least in commercial applications.

Open-sourcing the MuJoCo physics simulator for robotics

Related Stories