Agibot 출시 Genie Studio: 최초 All-에서-One Platform 위한 구현된 AI Development

On April 21, Agibot (ZhiYuan Robotics) launched Genie Studio, the industry’s first all-in-one development platform for embodied intelligence. Designed to address critical challenges like limited data access, high training costs, and complex engineering barriers, Genie Studio simplifies the full pipeline from data collection to model deployment—enabling developers to “train smarter robot brains” more efficiently than ever before.

Genie Studio features an end-to-end toolchain covering four key functions: data collection, model training, simulation evaluation, and model deployment. Its data module supports large-scale multimodal collection with visual annotation tools, robust dataset management, and compatibility with real-world and simulated robot bodies. It comes preloaded with millions of industry-verified data points across 100+ scenarios in sectors like industrial automation and smart homes.

The model training module serves as a “factory for embodied AI,” supporting popular open-source models (including Agibot GO-1, RDT, PI0, OpenVLA, and GR00T). It enables rapid scenario adaptation and dramatically reduces compute time with optimized GPU scheduling via Kubernetes and Volcano.

The simulation module offers a rich 3D asset library for testing algorithms safely and at scale. Developers can simulate complex environments, auto-generate training tasks, and validate models before real-world deployment. Local simulation tools will also be open-sourced to users in the Genie Studio and Agibot World ecosystem.

For deployment, Genie Studio offers GDK (Genie Development Kit) through Genie Store, enabling plug-and-play model injection into robots. It includes an integrated application management platform for updates, diagnostics, and optimization—paving the way for industrial-scale deployment and lifecycle management of embodied AI solutions.