PyBullet and “Sim-to-Real: Learning Agile Locomotion For Quadruped Robots”

PyBullet is receiving regular updates, you can see the latest version here: https://pypi.org/project/pybullet
Installation and update is simple:
pip install -U pybullet

Check out the PyBullet Quickstart Guide and clone the github repository for more PyBullet examples and OpenAI Gym environments.

A while ago, Our RSS 2018 paper “Sim-to-Real: Learning Agile Locomotion For Quadruped Robots” is accepted! (with Jie Tan, Tingnan Zhang, Erwin Coumans, Atil Iscen, Yunfei Bai, Danijar Hafner, Steven Bohez, Vincent Vanhoucke).

See also the video and paper on Arxiv.

Erwin @ twitter

Server Migration

The website and forum of Bullet just migrated to a new server. Things are under construction and http://pybullet.org and http://bulletphysics.org will be merged into one.

 

Bullet 2.87 with pybullet robotics Reinforcement Learning environments

Bullet 2.87 has improved support for robotics, reinforcement learning and VR. In particular, see the “Reinforcement Learning” section in the pybullet quickstart guide at http://pybullet.org . There are also preliminary C# bindings to allow the use of pybullet inside Unity 3D for robotics and reinforcement learning. In addition, vectorunit Beach Buggy Racing using Bullet has been released for the Nintendo Switch!

beach-buggy-racing

You can download the release from https://github.com/bulletphysics/bullet3/releases
Here are some videos of some Bullet reinforcement learning environments trained using TensorFlow Agents PPO:


See also KUKA grasping
and pybullet Ant

Bullet 2.86 with pybullet for robotics, deep learning, VR and haptics

The Bullet 2.86 has improved Python bindings, pybullet, for robotics, machine learning and VR, see the pybullet quickstart guide.

Furthermore, the PGS LCP constraint solver has a new option to terminate as soon as the residual (error) is below a specified tolerance (instead of terminating after a fixed number of iterations). There is preliminary support to load some MuJoCo MJCF xml files (see data/mjcf), and haptic experiments with a VR glove. Get the latest release from github here.App_SharedMemoryPhysics_VR_vs20 2017-01-26 10-12-45-16

[youtube 0JC-yukK-jo].C2guE9TUcAAeMOw

 

 

Bullet 2.85 released : pybullet and Virtual Reality support for HTC Vive and Oculus Rift

bullet_pybullet_vrWe have been making a lot of progress in higher quality physics simulation for robotics, games and visual effects. To make our physics simulation easier to use, especially for roboticist and machine learning experts, we created Python bindings, see examples/pybullet. In addition, we added Virtual Reality support for HTC Vive and Oculus Rift using the openvr sdk. See attached youtube movie. Updated documentation will be added soon, as well as possible show-stopper bug-fixes, so the actual release tag may bump up to 2.85.x. Download the release from github here.

[youtube VMJyZtHQL50].

Bullet 2.83 released and upcoming SIGGRAPH 2015 course

bullet2.83

The new Bullet Physics SDK 2.83 is available from github. The biggest change is the new example browser using OpenGL 3+. For more changes and features, see the docs/BulletQuickstart.pdf as part of the release. For more information and download link, see http://www.bulletphysics.org/Bullet/phpBB3/viewtopic.php?f=18&t=10527

Also, our proposal for a course on Bullet got accepted for the upcoming SIGGRAPH 2015 conference in Los Angeles.

Tuesday, 11 August 3:45 pm – 5:15 pm, Los Angeles Convention Center, Room 404AB

UPDATE: here are the slide decks:

3:45-4.15 pm
Introduction to rigid body pipeline, collision detection

4:15-4:45 pm
Advances in constraint solving, Featherstone Articulated Body Algorithm

4:45-5.15 pm
Acceleration of the full collision detection and constraint solver on GPU

Scientific and Technical Academy Award for the development of Bullet Physics!

The Academy of Motion Picture Arts and Sciences today announced that 21 scientific and technical achievements represented by 58 individual award recipients will be honored at its annual Scientific and Technical Awards Presentation on Saturday, February 7, at the Beverly Wilshire in Beverly Hills.

“To Erwin Coumans for the development of the Bullet physics library, and to Nafees Bin Zafar and Stephen Marshall for the separate development of two large-scale destruction simulation systems based on Bullet.

These pioneering systems demonstrated that large numbers of constrained rigid bodies could be used to animate visually complex, believable destruction effects with minimal simulation time.”

Thanks to all Bullet contributors and users!
See https://www.oscars.org/news/21-scientific-and-technical-achievements-be-honored-academy-awardsr

Bullet used in NASA Tensegrity Robotics Toolkit, book Multithreading for Visual Effects

multithreading

nasa_bullet

Nasa is using Bullet in their new open source Tensegrity Robotics Toolkit. You can find more information and video link here: http://bulletphysics.org/Bullet/phpBB3/viewtopic.php?f=17&t=9978

The new book Multithreading for Visual Effects includes a chapter on the OpenCL optimizations for upcoming Bullet 3.x. Other chapters include multithreading development experiences from OpenSubDiv, Houdini, Pixar Presto and Dreamworks Fluids and LibEE. You can get it at the publisher AK Peters/CRC Press or at Amazon.

Development on upcoming Bullet 2.83 and Bullet 3.x is making good progress, hopefully an update follows soon.

Bullet moves to github and Erwin Coumans joins Google!

GoogleBostonDynamics Development of the open source Bullet Physics SDK continues at http://github.com/bulletphysics/bullet3 . All the open issues have been moved from the googlecode repository to github with links between old and new issues. There will be a Bullet 2.83 release using the github repository very soon, it is in alpha stage now. In 2014 we will be moving to Bullet 3.x and the unstable Bullet 3.x code is already included in Bullet 2.83.
Other news is that recently I joined Google to work on the robotics project!

Bullet 2.82 released: Featherstone and direct MLCP solvers. Poser 3D, Riptide GP2 using Bullet.

The new Bullet 2.82 SDK is available for download. It allows for higher quality physics simulation, suitable for robotics, using the Featherstone articulated body algorithm. The release also introduces a new Mixed Linear Complementarity Problem (MLCP) solver interface, with various direct solver implementations. Read more here. Poser 3D using Bullet
Riptide GP2 using Bullet

Our focus is now on integrating all Bullet 2.x features into the upcoming Bullet 3.x SDK. You can learn more about its progress in our SIGGRAPH course notes on GPU rigid body simulation at the Multithreading and VFX website.

Ralf Knoesel shared the good news that Riptide GP2, the latest iOS/Android game from Vector Unit, uses Bullet for collision detection and rigid body simulation. See http://www.bulletphysics.org/Bullet/phpBB3/viewtopic.php?f=17&t=9324 for more information

We have been working with Larry Weinberg and his crew to integrate Bullet soft body and rigid body into Poser 3D. Create rigid simulations with constraints to build complex mechanical interactions. Paint softbody constraint weights to animate. Add jiggle and bounce to any prop or character. Use the Live Simulation mode to preview dynamics in realtime, or calculate simulations to include in rendered animations. See for more information on Bullet and Poser 3D here.