Axis Robotics has released Axis Sim Dataset V1, one of the largest open-source simulation datasets for Franka arm manipulation. The dataset contains over 50,000 human-teleoperated simulation trajectories across 207 manipulation tasks and 60,000+ scene variants on a simulated Franka Research 3 arm.
The dataset has become the most downloaded open-source simulation Franka manipulation dataset on Hugging Face, with 160,000+ downloads. In benchmarks, continual pretraining on V1 improved performance on the LIBERO-Plus benchmark, lifting success rates from 83.9% to 88.8% and outperforming a volume-matched RoboCasa365 baseline by 37.3%.
Data Collection and Methodology
The trajectories span pick-and-place, stacking, pouring, articulated-object manipulation, and tool use, collected through Axis's browser-based teleoperation platform, Axis Hub, by a distributed crowd rather than a single expert team. The dataset was built with researchers from UC Berkeley, Johns Hopkins, the University of Michigan, and other institutions.
Performance improvements appeared consistently as pretraining data scaled from 25% to 100% of the dataset, with the largest gains under camera, sensor-noise, and layout perturbations.
Broader Data Infrastructure
The dataset is part of a larger data engine spanning four operational lines: simulation with over 200,000 distributed contributors producing 4.7M+ trajectories across 13 embodiments; egocentric capture with 1,000+ full-time collectors capturing 200,000+ hours of first-person activity and growing by 4,000+ hours daily; loco-manipulation with 500+ hours on real humanoids; and human-gated correction targeting deployment edge cases.
Commercial Applications
Axis works with robot embodiment companies to build customized, embodiment-specific data pipelines. As Booster Robotics' first sim-data partner, Axis created a Booster-specific model prior from 42,000+ simulation episodes, reaching 87.5% success with 30 real-robot demos versus 37.5% for an out-of-the-box baseline.
The company raised $12 million in seed funding led by Hack VC, with participation from Nomad Capital, Pi Network Ventures, and 10K Ventures. V2 development is underway, scaling to 1.2 million trajectories across 1,200 tasks.
