Capture • Dataset • Evaluation

AXIS

A Growable Community-Driven Data Engine
for Scalable Robot Manipulation

Axis Robotics Team

AXIS turns accessible browser-based teleoperation into a continuously growing, training-ready robot manipulation dataset. It combines task generation, community demonstrations, automated validation, trajectory refinement, IsaacSim augmentation, and fixed-protocol policy evaluation.

[paper] [Dataset (Hugging Face)] [Platform] Dataset Browser
917,313Trajectories counted
1056h 32mEstimated duration
1,006Tasks counted
FR3Robot embodiment
AXIS system overview showing task generation, teleoperation, data cleaning, augmentation, policy learning, and real-world validation.

AXIS system overview: task generation, browser-based teleoperation, offline data processing, simulation augmentation, policy learning, and real-world validation.

Abstract.

Robot manipulation policies require diverse, high-quality demonstrations, but existing data pipelines are hard to scale because they often rely on specialized hardware, centralized operators, or fixed task suites. AXIS is a growable community-driven data engine and benchmark for scalable manipulation learning.

AXIS collects demonstrations through browser-based MuJoCo-WASM teleoperation, generates and validates new tasks automatically, and converts community demonstrations into training-ready data through success checking, filtering, smoothing, resampling, and IsaacSim-based visual and physics augmentation.

To make growth measurable, AXIS organizes data into task snapshots and evaluates policies with a fixed held-out protocol. The current experiments compare representative vision-language-action baselines and study scaling behavior across AXIS-25%, AXIS-50%, and AXIS-100%.

Scalable Robot Data Collection.

Collection stays lightweight in the browser; validation, rendering, augmentation, and learning run on the backend.

The AXIS collection layer separates latency-sensitive teleoperation from compute-heavy data processing. Contributors operate a Franka Research 3 arm in a MuJoCo-WASM browser frontend with commodity input devices, while uploaded trajectories are routed to backend machines for validation, cleaning, replay, rendering, augmentation, and export.

Data engine stages

  1. 01 Task generation and asset normalization
  2. 02 Web-based MuJoCo-WASM teleoperation
  3. 03 Success checking and trajectory upload
  4. 04 Filtering, hesitation removal, smoothing, and resampling
  5. 05 IsaacSim visual and physics augmentation
  6. 06 VLA training, fixed-protocol evaluation, and real-world rollout

Browser teleoperation

Contributors collect demonstrations without local simulator installation, GPUs, or specialized robot hardware.

Unified trajectory format

Each episode stores task metadata, embodiment, simulator version, observations, states, actions, and success information.

Quality refinement

Backend validation removes corrupted records, failed tasks, idle segments, jitter, and nonphysical discontinuities.

Realistic augmentation

Scene, camera, lighting, texture, pose, friction, mass, and dynamics variations expand the training distribution.

Browser-based teleoperation sessions showing remote task execution through the AXIS collection interface.
The AXIS Franka Dataset.

AXIS is a tabletop manipulation dataset built around a Franka Research 3 robot with a parallel-jaw gripper. Each task includes a language instruction, a parameterized simulation scene, task assets, and a structured success checker. Each trajectory provides synchronized robot states, object states, robot actions, task metadata, success labels, and multi-view RGB-D observations.

Dataset snapshot

Tasks
1,006
Trajectories
917,313
Duration
1056h 32m
Robot
Franka Research 3
Cameras
Third-view + wrist RGB-D
Evaluation
Fixed held-out protocol
AXIS dataset overview showing dataset scale, trajectory schema, augmentation, data growth, and task distribution.
Dataset overview covering scale, trajectory schema, validation and refinement, augmentation, growth snapshots, and scene/skill distribution.
Browser collection workflow for a manipulation task, including on-screen controls and simulator feedback.
Trajectory refinement

Raw community demonstrations may contain hesitation, jitter, low-frequency sampling artifacts, or invalid transitions. AXIS filters and refines these demonstrations before converting them into policy-learning data.

Effect of data refinement on teleoperation trajectory quality.
Data Version Sampling Rate Mean Acceleration Mean Jerk Replay Success
Raw Teleoperation 5.0 Hz 1.3539 11.5899 100.0%
Smoothed 5.0 Hz 0.6382 2.9160 91.4%
Smoothed + Resampled 20 Hz 0.4885 2.2243 86.2%
Augmentation diversity across example AXIS manipulation tasks.
Augmentation diversity across example tasks, covering randomized appearance, lighting, environments, and object poses.
A living benchmark

Instead of treating the dataset as a one-time release, AXIS organizes training data into progressively larger snapshots. AXIS-25%, AXIS-50%, AXIS-100%, and future versions preserve shared data format, task definitions, success checkers, rollout budgets, and evaluation tasks, so scaling studies can isolate the effect of training coverage.

AXIS-25%low-volume scaling point
AXIS-50%mid-scale task coverage
AXIS-100%current full snapshot
AXIS-...future growable releases
Grid of augmented tabletop simulation scenes with varied backgrounds and object layouts.
Additional augmented tabletop scenes with varied backgrounds, lighting, surface textures, object layouts, and camera viewpoints.
Simulation scenes spanning kitchen, tabletop, mug, and object manipulation tasks.
Scene and object diversity across simulated Franka manipulation tasks.

What the dataset contains

language instruction task assets success checker robot states object states robot actions third-view RGB-D wrist RGB-D metadata failure labels
Simulation demos
Same task with different augmentation settings.
Different task setup with varied scene assets and initial conditions.
Experiments.

The experiments evaluate whether AXIS pretraining improves downstream LIBERO-Plus robustness for π0.5, whether the improvement scales with AXIS data volume, and which perturbation axes benefit most from AXIS-style augmentation and task diversity.

Headline result. π0.5 + AXIS-100% reaches 79.4 overall LIBERO-Plus success, compared with 66.5 for vanilla π0.5 and 57.5 for a RoboCasa-matched simulation baseline.
Main result on LIBERO-Plus.
Pretraining # demos Overall ↑ Cam. Light Noise B.G. Layout Lang. Robot
π0.5 vanilla 0 66.5 50.3 90.6 61.8 87.7 61.7 71.9 54.8
π0.5 + AXIS-25% 0.25 NAXIS 72.0 56.1 91.8 68.1 89.1 72.1 76.0 61.3
π0.5 + AXIS-50% 0.50 NAXIS 75.1 59.7 92.5 72.5 89.6 77.9 78.3 65.1
π0.5 + AXIS-100% (ours) NAXIS 79.4 65.5 93.1 78.5 90.4 84.9 81.1 70.2
π0.5 + RoboCasa-matched NAXIS 57.5 35.2 79.5 63.2 81.7 68.0 49.2 39.4
Per-perturbation robustness

Layout

+23.2

Largest gain over vanilla, matching AXIS spatial randomization.

Sensor Noise

+16.7

Robustness improves under photometric observation shifts.

Robot Pose

+15.4

AXIS helps beyond explicitly randomized perturbation axes.

Camera

+15.2

Viewpoint variation transfers to LIBERO-Plus camera shifts.

AXIS-100% improvements by perturbation axis.
Axis π0.5 vanilla + AXIS-25% + AXIS-100% + RoboCasa-m. ∆van AXIS−RC
Camera 50.3 56.1 65.5 35.2 +15.2 +30.3
Light 90.6 91.8 93.1 79.5 +2.5 +13.6
Sensor Noise 61.8 68.1 78.5 63.2 +16.7 +15.3
Background 87.7 89.1 90.4 81.7 +2.7 +8.7
Layout 61.7 72.1 84.9 68.0 +23.2 +16.9
Language 71.9 76.0 81.1 49.2 +9.2 +31.9
Robot 54.8 61.3 70.2 39.4 +15.4 +30.8
Overall 66.5 72.0 79.4 57.5 +12.9 +21.9
Real-world rollouts
Third-view rollout with tabletop grocery objects.
Third-view rollout under the same real-world setup.
Wrist-view rollout with close-up manipulation observations.
Wrist-view rollout with top-down task observations.
Real-world Franka rollout videos under third-view and wrist-view observations.
BibTeX.
BibTeX 
@misc{axis2026growable,
  title        = {AXIS: A Growable Community-Driven Data Engine for Scalable Robot Manipulation},
  author       = {{Axis Robotics Team}},
  year         = {2026},
  note         = {CoRL 2026 submission}
}