{"id":7056839,"date":"2026-09-04T09:21:17","date_gmt":"2026-09-04T09:21:17","guid":{"rendered":"https:\/\/som2nypost.com\/uncategorized\/axis-robotics-open-sources-one-of-the-largest-franka-arm-simulation-datasets-for-physical-ai\/"},"modified":"2026-09-04T09:21:17","modified_gmt":"2026-09-04T09:21:17","slug":"axis-robotics-open-sources-one-of-the-largest-franka-arm-simulation-datasets-for-physical-ai","status":"publish","type":"post","link":"https:\/\/som2nypost.com\/?p=7056839","title":{"rendered":"Axis Robotics Open-Sources One of the Largest Franka Arm Simulation Datasets for Physical AI"},"content":{"rendered":"


\n<\/p>\n

\n

Axis Robotics has released <\/span>Axis Sim Dataset V1<\/span><\/a>, one of the largest open-source simulation datasets for Franka arm manipulation, with the full dataset, training code, and benchmarks publicly available. V1 is built from more than 50,000 human-teleoperated simulation trajectories across 207 manipulation tasks and 60,000+ scene variants on a simulated Franka Research 3 arm.<\/span><\/p>\n

This dataset drew over 160,000 downloads, making it the most downloaded open-source simulation Franka manipulation dataset on Hugging Face. In benchmarks, continual pretraining on V1 lifted \u03c00.5 and beat a volume-matched RoboCasa baseline, with every result open and verifiable.<\/span><\/p>\n

\"\"<\/p>\n

Axis Robotics is building the ultimate compounding data engine for Physical AI, a vertically integrated system spanning large-scale simulation, egocentric real-world capture, humanoid loco-manipulation, and human-gated DAgger post-training. The company raised $12 million in seed funding led by Hack VC, with participation from Nomad Capital, Pi Network Ventures, 10K Ventures, and angel investors.<\/span><\/p>\n

A Bet Against \u201cClean Data Only\u201d<\/b><\/h2>\n

A common assumption in robotics is that demonstrations must be near-optimal to begin with \u2014 filter down to expert trajectories, standardize the setup, and discard anything noisy before it is safe to imitate. Axis\u2019s thesis runs the other way: data quality lives at the distribution level, not the single trajectory. When a large and diverse enough crowd produces noisy, suboptimal trajectories and their errors are uncorrelated, the noise averages out and a working policy survives during training.<\/span><\/p>\n

Axis Sim Dataset V1 puts that thesis to a public test. Its trajectories span pick-and-place, stacking, pouring, articulated-object manipulation, and tool use, all collected through Axis\u2019s 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.<\/span><\/p>\n

Results That Scale<\/b><\/h2>\n

On LIBERO-Plus, continual pretraining on V1 lifts \u03c00.5 from 83.9% to 88.8% success and outperforms a volume-matched RoboCasa365 baseline by 37.3%. Performance improves consistently as pretraining data scales from 25% to 100% of the dataset, with no saturation in sight, evidence that the gains come from diversity and coverage rather than a one-off bump. The largest improvements appear under camera, sensor-noise, and layout perturbations, the exact axes Axis randomizes during generation.<\/span><\/p>\n

\"\"<\/p>\n

The team says V2 is already underway, scaling to 1.2 million trajectories across 1,200 tasks, with cross-embodiment generalization and results across multiple VLA models showing that suboptimal simulation data trains robust policies.<\/span><\/p>\n

The Engine Behind the Dataset<\/b><\/h2>\n

The dataset is one output of a larger, actively compounding data engine. Where a traditional data vendor collects to a fixed spec and stops, Axis uses model performance and failure cases to determine what should be collected next, so every training round informs the next. That engine runs on a hybrid strategy across four data lines, and all four now run at scale:<\/span><\/p>\n