The AIRBOT Play is a compact 6-degree-of-freedom collaborative robotic arm designed for embodied AI research, robot learning, automation, and educational applications. Combining a lightweight structure, high payload-to-weight ratio, and open software ecosystem, it serves as an ideal platform for developing intelligent manipulation algorithms and robotic applications.

Supporting multiple control modes, force compensation, and seamless integration with wheeled robots, quadruped robots, UAVs, and desktop workstations, AIRBOT Play enables flexible deployment across both research and industrial environments.

Key Features

Lightweight & Portable Design

Weighing only 3.78 kg, the robotic arm offers excellent portability while maintaining a 1.5 kg continuous payload, making it suitable for mobile robotic platforms and laboratory environments.

AI-Powered Intelligent Manipulation

Integrated with large language model (LLM) capabilities, AIRBOT Play can understand natural language commands, automatically plan manipulation tasks, and support intelligent embodied AI applications. It also supports scene scanning and simulation for accelerating robot learning workflows.

Open Development Platform

Developers can quickly build custom robotics applications using:

• Python SDK
• C++ SDK
• ROS2 Support
• Position Control
• Velocity Control
• Torque Control

The open architecture supports rapid secondary development and integration into various robotic systems.

Wide Compatibility

The robotic arm can be installed on:

• Desktop Platforms
• Wheeled Mobile Robots
• Quadruped Robots
• Wheeled-Leg Robots
• UAV Platforms
• Custom Automation Equipment

Technical Specifications

Specifications are compiled from the product brochure.

Hardware Expansion

AIRBOT Play supports a wide range of peripherals and development accessories, including:

• RGB Cameras
• RGB-D Cameras
• Mini PCs
• Embedded GPU/BPU Computing Modules
• Electric Grippers
• Teaching Grippers
• Dexterous Hands
• Custom End Effectors

Multiple hardware interfaces, including USB, GPIO, UART, I²C, CAN, and RS485, provide excellent flexibility for research and industrial development.

Software Ecosystem

The platform provides a complete software toolchain covering the entire embodied AI workflow:

• Data Collection
• Data Annotation
• Model Training
• Model Evaluation
• Robot Deployment
• High-Fidelity Simulation
• Sim-to-Real Development

Simulation environments enable rapid algorithm verification while improving research efficiency.

Applications

AIRBOT Play is ideal for:

• Embodied AI Research
• Robot Learning
• Reinforcement Learning
• Imitation Learning
• AI Manipulation Research
• Robotics Education
• University Laboratories
• Industrial Automation
• Mobile Manipulation
• Vision-Guided Robotics
• Human-Robot Interaction
• Autonomous Service Robots

Advantages

• Lightweight collaborative robot arm
• High payload-to-weight ratio
• ±0.1 mm repeatability
• Supports Python, C++, and ROS2
• LLM-enabled intelligent manipulation
• Open SDK for rapid development
• Rich hardware expansion interfaces
• Compatible with multiple mobile robot platforms
• Designed for embodied AI research and education

Available Robotic Platforms

To support different research and deployment requirements, AIRBOT Play is also available as part of several integrated robotic platforms:

• MMK2 Intelligent Mobile Dual-Arm Platform – Mobile dual-arm robot for multimodal embodied AI research and data collection.
• TOK4 Teleoperation Kit – Mobile teleoperation platform supporting imitation learning and remote manipulation.
• Quadruped Mobile Manipulation Platform – Combines a quadruped robot with AIRBOT Play for navigation and manipulation in complex terrain.
• Wheeled-Leg Mobile Manipulation Platform – High-mobility platform with active suspension for logistics, inspection, and autonomous manipulation.