MuJoCo: A physics engine for model-based control

By E. Todorov, Tom Erez, Yuval Tassa. 2012

7,536 citations recorded by the source corpus.

Subfields: Computational Theory and Mathematics, Control and Systems Engineering, Mechanical Engineering, Robotics & Control

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Control and Systems Engineering

In Mechanical Engineering

In Robotics & Control

Citations in the corpus

Proximal Policy Optimization Algorithms (2017, methodology): “Namely, we used 7 simulated robotics tasks 2 implemented in OpenAI Gym [Bro+16], which use the MuJoCo [TET12] physics engine.”
Trust Region Policy Optimization (2015, methodology): “We conducted the robotic locomotion experiments using the MuJoCo simulator (Todorov et al., 2012).”
Continuous control with deep reinforcement learning (2015, positive): “These environments were simulated using MuJoCo [13].”
Experience Grounds Language (2020, background): “The last few years have seen massive advances in both high ﬁdelity simulators for robotics (Todorov et al., 2012; Coumans and Bai, 2016–2019; NVIDIA, 2019; Xiang et al., 2020) and the cost and availability of commodity hardware (Fitzgerald, 2013; Campeau-Lecours et al., 2019; Murali et al., 2019).”
Domain randomization for transferring deep neural networks from simulation to the real world (2017, methodology): “We render images using the MuJoCo Physics Engine’s [45] built-in renderer.”
Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks (2017, methodology): “complex deep RL problems, we also study adaptation on high-dimensional locomotion tasks with the MuJoCo simulator (Todorov et al., 2012).”
Deep Reinforcement Learning from Human Preferences (2017, methodology): “The first tasks we consider are eight simulated robotics tasks, implemented in MuJoCo (Todorov et al., 2012), and included in OpenAI Gym (Brockman et al.”
Emergence of Locomotion Behaviours in Rich Environments (2017, methodology): “All environments rely on the Mujoco physics engine [21].”
Sample Efficient Actor-Critic with Experience Replay (2016, methodology): “We evaluate our algorithms on 6 continuous control tasks, all of which are simulated using the MuJoCo physics engine (Todorov et al., 2012).”
OpenAI Gym (2016, methodology): “These tasks use the MuJoCo physics engine, which was designed for fast and accurate robot simulation [14].”
Benchmarking Deep Reinforcement Learning for Continuous Control (2016, methodology): “Tasks with more complicated dynamics, such as locomotion, are implemented using MuJoCo (Todorov et al., 2012), a 3D physics simulator with better modeling of contacts.”
High-Dimensional Continuous Control Using Generalized Advantage Estimation (2015, methodology): “The simulated robot tasks were simulated using the MuJoCo physics engine (Todorov et al., 2012).”
Memory-based control with recurrent neural networks (2015, methodology): “All physical control problems except the simulated water maze (section 4.3) were simulated in MuJoCo [28].”
Learning Continuous Control Policies by Stochastic Value Gradients (2015, background): “2), we show that SVG(1) can be applied to several complicated, multidimensional physics environments involving contact dynamics (Figure 1) in the MuJoCo simulator [28].”
End-to-End Training of Deep Visuomotor Policies (2015, methodology): “resent a detailed summary of the experimental setup for our simulated and real-world experiments. B.1 Simulated Experiment Details All of the simulated experiments used the MuJoCo simulation package (Todorov et al., 2012), with simulated frictional contacts and torque motors at the joints used for a”
Programmable Agents (2017, methodology): “The programmable reaching environment is implemented with the MuJoCo [34] physics engine, and hence the objects are subject to friction, contact forces, gravity, etc.”
A simple neural network module for relational reasoning (2017, methodology): “We developed a dataset of simulated physical mass-spring systems using the MuJoCo physics engine [21].”
Pre-Trained Language Models for Interactive Decision-Making (2022, methodology): “Policy learning and LM. Traditional policy learning methods, such as PPO [37], DQN [27], DDPG [21], A3C [26], perform well on playing tasks on Atari, OpenAI gym [5], and MuJoCo [41].”
ALFWorld: Aligning Text and Embodied Environments for Interactive Learning (2020, background): “While ﬁelds such as robotic control use simulators like MuJoCo (Todorov et al., 2012) to provide inﬁnite data through interaction, there has been no analogous mechanism – short of hiring a human around the clock – for providing linguistic feedback and annotations to an embodied agent.”
Measuring the Intrinsic Dimension of Objective Landscapes (2018, methodology): “5, we show results of ES on three tasks: InvertedPendulum−v1, Humanoid−v1 in MuJoCo (Todorov et al., 2012), and Pong−v0 in Atari.”