Axis Robotics has launched Axis Sim Dataset V1, an open-source collection for Franka Research 3 arm manipulation. It contains over 50,000 human-teleoperated trajectories covering 207 tasks and 60,000 scene variants. The dataset has surpassed 160,000 downloads on Hugging Face.
The company secured $12 million in a seed funding round led by Hack VC. Participants included Nomad Capital, Pi Network Ventures, 10K Ventures, and several angel investors. The startup was founded by researchers from institutions including UC Berkeley, CMU, and Georgia Tech.
Benchmarks show that pretraining on V1 improved the success rate of the π0.5 model from 83.9% to 88.8% on LIBERO-Plus. The performance outperformed the RoboCasa365 baseline by 37.3%. Axis Robotics is currently developing V2, which aims to scale to 1.2 million trajectories.
Axis Robotics operates a data engine that records tasks and trajectories on-chain via the Base network. Its simulation platform, Axis Hub, involves over 200,000 contributors. The company also manages a network of 1,000 collectors for egocentric data, currently totaling over 200,000 hours of footage.
The firm collaborates with partners such as Booster Robotics, Feagine Robotics, Manycore Tech, Dexmal, Lotus Cars, and Geely Auto. It also provides data for on-chain networks including BitRobot on Solana and OpenRoboto on the Bittensor blockchain.
