Embodied AI is rapidly advancing into the real world. From basic grasping to complex manipulation, robotic capabilities continue to expand, placing higher demands on the performance, efficiency, and ease of use of development platforms.

How to make the development workflow more efficient and convenient — from hardware setup and data collection to model deployment — has become a critical challenge in embodied intelligence research and development.

Today, Galaxea's LEMO Dual-Arm Development Platform is officially available for order!

LEMO is an all-in-one platform that integrates high-performance dual-arm hardware, isomorphic teleoperation, multimodal data collection, and an open development toolchain. It covers the complete development pipeline — from data collection and model training to on-robot deployment and validation — providing a one-stop platform for embodied AI research and application development.

Launch price: RMB 49,999. Orders are now open.

LEMO — Build faster. Create more.

High-Performance Dual-Arm Hardware, Expanding the Boundaries of Robotic Manipulation

LEMO adopts a modular desktop workstation design, integrating dual A1Z robotic arms, G1Z grippers, a multi-view synchronous vision system, and a CCU controller. It combines dual-arm operation, visual perception, and motion control capabilities in one system, reducing the developer's investment in hardware selection, system integration, and debugging.

Core hardware highlights — LEMO offers the following advantages:

• 14 DOF dual-arm collaboration: Equipped with dual A1Z 7-DOF robotic arms, supporting flexible dual-arm coordinated operation.
• 3 kg rated payload / 5 kg peak payload: Balances payload capacity with operational flexibility, expanding the range of robotic objects and task types. (Note: Payload specifications do not include end effectors.)
• ±0.1 mm repeat positioning accuracy: Meets experimental requirements for precision tasks such as fine manipulation, object grasping, and assembly.
• Modular workstation design: Supports adjusting the dual-arm spacing according to experimental needs, adapting to different task configurations.
• Flexible compute expansion: Supports integration with external compute devices such as Orin, PC, and Mac, meeting the needs of different development environments.
• From basic grasping to complex dual-arm operation, LEMO provides a high-performance, flexible, and scalable hardware foundation for diverse robotic tasks.

Integrated Teleoperation and Multimodal Data Collection, R&D Efficiency Upgraded

LEMO integrates isomorphic teleoperation and multimodal data collection on the same platform, helping developers efficiently complete operation teaching and data acquisition.

In terms of teleoperation and data collection, LEMO supports:

• Isomorphic master-slave teleoperation: Paired with A1Z-T for smooth, responsive fine manipulation and teaching.
• VR takeover: Provides flexible operation modes to adapt to different task data collection requirements.
• Three-channel synchronous vision collection: Integrates dual wrist cameras and a top camera to acquire multi-view operational data.
• Precise multimodal data alignment: Supports synchronous collection of vision and robotic arm states, timestamp alignment, and standardized output.
• From teleoperation teaching to high-quality data collection, LEMO provides a flexible experimental platform for imitation learning, reinforcement learning, dual-arm coordinated operation, and embodied AI model research.

Open Development Toolchain, Supporting Model Training to Real-Robot Validation

LEMO provides open development tools and interfaces, supporting developers in flexibly integrating algorithms and deploying models, accelerating embodied AI R&D and application validation.

In terms of development tools and ecosystem support, LEMO provides:

• Rich development resources: Provides SDK, URDF, 3D models, and open interfaces to support secondary development.
• Model deployment and real-robot validation: Supports algorithm deployment, task execution, and performance evaluation.
• Diverse algorithm exploration: Supports research on imitation learning, real-robot reinforcement learning, and other technologies, helping continuously optimize robotic manipulation policies.
• LEMO helps developers reduce the cost of building underlying systems, allowing them to focus more energy on algorithm innovation and application development.
• From embodied AI foundation models and robotic hardware platforms to open development tools, Galaxea continues to build an embodied intelligence technology ecosystem for the real world.
• With the official launch of LEMO, we look forward to more developers joining the exploration — turning new algorithms and new ideas into reality on real robots.
• LEMO is ready. Build faster. Create more.