Figure Unveils Index, a Global Data Engine for Training General-Purpose Robots
Index is already the largest real-world dataset ever built for training general-purpose humanoid robots.
Index is already the largest real-world dataset ever built for training general-purpose humanoid robots.
German Solar Power selected the Bittensor subnet to build autonomous drone monitoring for solar parks in Greece.
Axis will contribute more than 3 million multimodal trajectories and support OpenRoboto's benchmarking stack as the Bittensor Subnet 80 robotics competition expands its data pipeline.
OpenRoboto is now live on mainnet, bringing a public benchmark and incentive system for improving robotics models to Bittensor.
Nepher Robotics runs Bittensor Subnet 49, where miners train Boston Dynamics Spot navigation policies in NVIDIA Isaac Sim, and validators score them in sandboxed containers, with only the top policy each cycle being paid.
SWARM (Subnet 124) is Bittensor's open benchmark for autonomous drone flight, training AI pilots with a depth camera instead of GPS or maps. Its model, SOTAPilot, hit a 95.34% success rate across six procedural worlds, then flew on real hardware via the open-source Langostino drone.