All posts by admin

Tiny differentiable simulator

We are developing a new differentiable simulator for robotics learning, called Tiny Differentiable Simulator, or TDS. The simulator allows for hybrid simulation with neural networks. It allows different automatic differentiation backends, for forward and reverse mode gradients. TDS can be trained using Deep Reinforcement Learning, or using Gradient based optimization (for example LFBGS). In addition, the simulator can be entirely run on CUDA for fast rollouts, in combination with Augmented Random Search. This allows for 1 million simulation steps per second.

Tiny Differentiable Simulator training a policy using ARS simulation in CUDA

TDS is used in a couple of research papers, such as the ICRA 2021 NeuralSim: Augmenting Differentiable Simulators with Neural Networks”, by Eric Heiden, David Millard, Erwin Coumans, Yizhou Sheng and Gaurav S. Sukhatme.

You can checkout the source code here:

PyBullet in a colab

You can directly try out PyBullet in your web browser, using Google Colab. !pip install pybullet takes about 15 seconds, since there is a precompiled Python wheel.

Here is an example training colab a PyBullet Gym environment using Stable Baselines PPO:

It is also possible to use hardware OpenGL3 rendering in a Colab, through EGL. Make sure to set the MESA environment variables. Here is an example colab.

PyBullet/Bullet Physics 3.05

Tagged a github release of Bullet Physics and PyBullet, both version 3.05. The release was used for our motion imitation research, and also includes various improvements for the finite-element-method (FEM) deformable simulation, by Xuchen Han and Chuyuan Fu .

See also the related repository with model-predictive-control (MPC) quadruped locomotion and the motion imitation research at

Assistive gym @ ICRA, 2020

Assistive Gym leverages PyBullet for physical human-robot interaction and assistive robotics. Assistive Gym currently supports four collaborative robots and six physically assistive tasks.
It also supports learning-based control algorithms, and includes models of human motion, human preferences, robot base pose optimization, and realistic pose-dependent human joint limits

Paper Link: Zackory Erickson, Vamsee Gangaram, Ariel Kapusta, C. Karen Liu, and Charles C. Kemp, “Assistive Gym: A Physics Simulation Framework for Assistive Robotics”, IEEE International Conference on Robotics and Automation (ICRA), 2020.

Learning Agile Robotic Locomotion Skills by Imitating Animals

In “Learning Agile Robotic Locomotion Skills by Imitating Animals”, we present a framework that takes a reference motion clip recorded from an animal (a dog, in this case) and uses RL to train a control policy that enables a robot to imitate the motion in the real world. By providing the system with different reference motions, we are able to train a quadruped robot to perform a diverse set of agile behaviors, ranging from fast walking gaits to dynamic hops and turns. The policies are trained primarily in simulation, and then transferred to the real world using a latent space adaptation technique that can efficiently adapt a policy using only a few minutes of data from the real robot. All simulations are performed using PyBullet.

See also and