Agibot X1

Overview

AgiBot X1 is a “bring-your-own-brain” humanoid: you supply an x86 PC or Jetson, it supplies clean, low-latency actuation and open tooling. 34 quasi-direct-drive joints deliver smooth gait and responsive upper-body motion, while the OmniPicker grippers handle common lab tasks up to 0.5 kg per hand. A modular, serviceable design—with neatly routed cabling and a CNC’d aluminium spine—keeps bench work simple. Open hardware files, firmware, and a Linux/ROS 2 stack make it easy to integrate RL pipelines or external LLMs. With ~2 hours of cordless testing and a developer-friendly price point, X1 is a practical platform for rapid iteration in humanoid research and education.

Key Features

• Height / Weight: 130 cm · 33 kg
• Top Speed: 3.6 km/h
• Runtime per Charge: ~2 h (swappable battery)
• Overall DOF: 34
• Hands / Fingers: 4-finger grippers (each lifts ~0.5 kg)
• Safety: Yes (human-safe operation)
• Verification: Not verified
• Compute: BYO host (x86 or NVIDIA Jetson); onboard DCU real-time controllers (1 kHz, 100 Mbps)
• OS / Middleware: Linux; ROS 2 / AimRT (open-source)
• Perception: Optional RGB-D camera
• Connectivity: Ethernet, FDCAN/CAN-FD, UART, GPIO, OTA firmware
• LLM/AI: DIY—hooks for external LLMs via ROS 2/AimRT
• Motors: PowerFlow quasi-direct-drive servo modules (R86/R52/L28)
• Gearing: Integrated low-ratio (<10:1) planetary reducers in servos
• Frame / Shell: Aluminium spine with composite outer panels
• Research labs and universities (embodied AI, locomotion, manipulation)
• Developer communities / startups prototyping skills and applications
• Education: hands-on courses in ROS 2, RL, and real-time control

Open-Source Research Platform

AgiBot launched the X1 as their dedicated research platform, distinct from their commercial A2 series. The X1's philosophy prioritizes hackability over polish — AgiBot provides full CAD files, assembly guides, component specifications, complete source code for locomotion, perception, and control, and access to the AimRT open-source robotics middleware. Researchers can modify core systems, integrate custom sensors, implement novel AI architectures directly on hardware, and reproduce published research. This level of openness goes significantly beyond typical SDK access offered by competitors.

Key Capabilities

• Fully open-source hardware documentation including CAD files and assembly guides
• AimRT framework — AgiBot's open-source robotics middleware, ROS-compatible
• 34 degrees of freedom for research-grade bipedal locomotion and manipulation
• PowerFlow servo actuators with integrated motor drivers and force feedback
• AGIBOT World Dataset access for embodied AI training
• Dual-arm configuration with 0.5 kg payload per arm
• Sensor suite: RGB cameras, LiDAR, force sensors, IMU
• ~2 hours autonomous operation
• Commercially available now