Datasetpublic

wujiss1/MoSim_Dataset

🗂️ MoSim Dataset Official release of the dataset from the paper:Neural Motion Simulator: Pushing the Limit of World Models in Reinforcement Learning This dataset contains sequential state-action trajectories for training and evaluating MoSim (Neural Motion Simulator) world models.All trajectories are collected from random policies in classical control and locomotion environments. 📦 Dataset Overview Format: .npz (NumPy compressed arrays) Contents:… See the full description on the dataset page: https://huggingface.co/datasets/wujiss1/MoSim_Dataset.

sourceHugging Facemitupdated 1y agoView on Hugging Face

0likes121downloads

Dataset Card

🗂️ MoSim Dataset

Official release of the dataset from the paper: [Neural Motion Simulator: Pushing the Limit of World Models in Reinforcement Learning](https://arxiv.org/pdf/2504.07095)

This dataset contains sequential state-action trajectories for training and evaluating MoSim (Neural Motion Simulator) world models. All trajectories are collected from random policies in classical control and locomotion environments.

📦 Dataset Overview

—Format: .npz (NumPy compressed arrays)
—Contents:
—*_random.npz: training episodes
—*_random_test.npz: test episodes
—Episode length: 1000 steps per episode

📊 Data Structure

Each .npz file contains:

Key	Shape	Description
`states`	(num_episodes, 1000, state_dim)	State at each timestep
`actions`	(num_episodes, 1000, action_dim)	Action applied at each timestep

State composition:

1.Joint DOF (articulated body)
2.Joint angles (radians)
3.Joint angular velocities (rad/s)
4.Root DOF (global free body)
5.Root position (x, y, z)
6.Root linear velocity (vx, vy, vz)
7.Root Orientation & Rotation
8.Root rotation quaternion (qx, qy, qz, qw)
9.Root angular velocity (wx, wy, wz)

⚡ For manipulation control tasks like Panda, only joint angles and velocities are provided. ⚡ For locomotion tasks like Humanoid or Go2, full root DOF and velocities are included.