REVIEW 21 cited by
Predictive Sampling: Real-time Behaviour Synthesis with MuJoCo
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Predictive Sampling: Real-time Behaviour Synthesis with MuJoCo
read the original abstract
We introduce MuJoCo MPC (MJPC), an open-source, interactive application and software framework for real-time predictive control, based on MuJoCo physics. MJPC allows the user to easily author and solve complex robotics tasks, and currently supports three shooting-based planners: derivative-based iLQG and Gradient Descent, and a simple derivative-free method we call Predictive Sampling. Predictive Sampling was designed as an elementary baseline, mostly for its pedagogical value, but turned out to be surprisingly competitive with the more established algorithms. This work does not present algorithmic advances, and instead, prioritises performant algorithms, simple code, and accessibility of model-based methods via intuitive and interactive software. MJPC is available at: github.com/deepmind/mujoco_mpc, a video summary can be viewed at: dpmd.ai/mjpc.
Forward citations
Cited by 21 Pith papers
-
Certified Gradient-Based Contact-Rich Manipulation via Smoothing-Error Reachable Tubes
A certified gradient-based method for contact-rich manipulation that quantifies smoothing-induced errors via set-valued discrepancies and incorporates them into analytical reachable sets for robust affine feedback policies.
-
On Surprising Effects of Risk-Aware Domain Randomization for Contact-Rich Sampling-based Predictive Control
Risk-aware domain randomization in contact-rich sampling-based predictive control reshapes the basin of attraction around contact-producing actions in the optimizer's effective cost landscape.
-
KinDER: A Physical Reasoning Benchmark for Robot Learning and Planning
KinDER is a new open-source benchmark that demonstrates substantial gaps in current robot learning and planning methods for handling physical constraints.
-
GPU-Accelerated Continuous-Time Successive Convexification for Contact-Implicit Legged Locomotion
ci-SCvx introduces integral cross-complementarity constraints to continuous-time SCP for contact-implicit legged locomotion, achieving faster solves and lower energy in MuJoCo validation than MPC baselines.
-
Approximating Global Contact-Implicit MPC via Sampling and Local Complementarity
A hybrid controller samples low-dimensional end-effector targets for a contact-free stage then runs local complementarity MPC at each sample to approximate global contact-implicit optimization.
-
ReKep: Spatio-Temporal Reasoning of Relational Keypoint Constraints for Robotic Manipulation
ReKep encodes robotic tasks as optimizable Python functions over 3D keypoints that are generated automatically from language and RGB-D input, enabling real-time hierarchical planning on single- and dual-arm platforms ...
-
Smooth Operator: A Real-Time Sampling-Based Algorithm for Kinematic Hand Retargeting
A sampling-based retargeter reduces hand teleoperation jitter and improves task success rates and operator workload compared to gradient-based baselines.
-
Physics-Informed Eikonal Caging for Whole-Arm Manipulation Planning
Reformulates caging as an eikonal minimum-time escape field, approximated via physics-informed neural network and embedded into whole-arm manipulation planning for improved robustness to contact-model mismatch.
-
Global Convergence of Sampling-Based Nonconvex Optimization through Diffusion-Style Smoothing
Recasts sampling-based nonconvex optimization as smoothed gradient descent to obtain non-asymptotic convergence guarantees and introduces the DIDA annealed algorithm that converges to the global optimum.
-
Tempered Sequential Monte Carlo for Trajectory and Policy Optimization with Differentiable Dynamics
Tempered sequential Monte Carlo samples efficiently from a temperature-annealed distribution over controller parameters to solve trajectory and policy optimization under differentiable dynamics.
-
Tempered Sequential Monte Carlo for Trajectory and Policy Optimization with Differentiable Dynamics
Tempered sequential Monte Carlo samples from a Boltzmann-tilted distribution over controllers to optimize trajectories and policies under differentiable dynamics.
-
Sumo: Dynamic and Generalizable Whole-Body Loco-Manipulation
Test-time steering of pre-trained whole-body policies via sample-based planning lets legged robots generalize dynamic loco-manipulation to varied heavy objects and tasks without additional training or tuning.
-
DynaRetarget: Dynamically-Feasible Retargeting using Sampling-Based Trajectory Optimization
DynaRetarget refines human kinematic motions into dynamically feasible humanoid trajectories using incremental sampling-based trajectory optimization, achieving higher success rates than prior methods on diverse objec...
-
GATO: GPU-Accelerated and Batched Trajectory Optimization for Scalable Edge Model Predictive Control
GATO is a new batched GPU trajectory optimization solver that achieves real-time MPC throughput with 18-21x speedups over CPU baselines for tens to low-hundreds of simultaneous solves.
-
RGB: RL Guided Whole-Body MPPI for Humanoid Control
RL policy biases MPPI sampling for humanoid control, enabling online correction of drift and new whole-body references in simulation.
-
Enabling Robust Cloth Manipulation via Inference-Time Simulator-in-the-Loop Refinement
Simulator-in-the-loop refinement with synthetic-data real-to-sim mapping and prior-guided MPPI planning yields higher success rates on real-robot cloth tasks than baselines.
-
TAM: Torque Adaptation Module for Robust Motion Transfer in Manipulation
TAM is a policy-agnostic torque adaptation module trained in randomized simulation that improves zero-shot real-robot performance on dynamic manipulation tasks compared to system identification and RMA baselines.
-
Learning Action-Conditional and Object-Centric Gaussian Splatting World Models for Rigid Objects
MRO-GWM learns action-conditional 3D dynamics of rigid objects via object-centric Gaussians and a transformer, evaluated on synthetic multi-object scenes and in simulation for non-prehensile manipulation.
-
PRIME: Physically-consistent Robotic Inertial and Motion Estimation for Legged and Humanoid Robots
PRIME is a MAP optimization framework that refines onboard kinematics into dynamically consistent trajectories for legged robots while jointly estimating contact forces and inertial parameters using differentiable smo...
-
Behavior Synthesis via Contact-Aware Fisher Information Maximization
Derives a contact-aware Fisher information measure to synthesize robot behaviors that maximize information-rich contacts for efficient object parameter learning.
-
Efficient On-policy Visual-RL via Stochastic Decoupled Policy Gradient
SDPG is a new on-policy visual RL algorithm that estimates gradients via stochastic perturbations of rollouts, achieving faster training and lower memory use than baselines on visual MuJoCo tasks while adding new robo...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.