The G1-D Ultimate is the mobile-chassis version of the G1-D platform. It adds a dual-wheel differential chassis with 360° on-spot rotation, a built-in 6-hour chassis battery, chassis LiDAR, dual depth cameras, and an auto-charging station. Built for teams that need full mobility alongside upper-body data capture and manipulation capability. Also available as the G1-D Standard with a fixed base.
Product Description
Unitree G1-D Ultimate — Wheeled Dual-Arm Humanoid Robot for AI Data Collection & Research
The Unitree G1-D Ultimate is the mobile-chassis version of Unitree's purpose-built dual-arm humanoid platform for embodied AI data collection, robot learning research, and manipulation development. It shares the same high-DOF upper body, multi-perspective camera array, and NVIDIA Jetson Orin processing power as the G1-D Standard — and adds a motorized dual-wheel differential chassis, built-in 6-hour battery, chassis LiDAR, dual depth cameras, collision and obstacle sensors, and an auto-charging station for teams that need full mobility alongside structured data capture.
With 19 degrees of freedom in the core body (expandable to 21–33 DOF depending on end effector selection), a 450mm vertical lifting stroke, and dual 7-DOF arms capable of 3kg payload at 0.45m reach, the G1-D Ultimate delivers the physical dexterity and mobile range needed to perform realistic manipulation tasks across dynamic, real-world environments — the kind of varied interaction data that fixed platforms and simulation alone can't replicate.
Choose from five end effector configurations — from a precision 2-finger force-controlled gripper to a 5-finger tactile dexterous hand — to match your specific data collection or research requirements. Also available as the G1-D Standard with a fixed base for controlled, stationary environments.
Why the G1-D Ultimate
Mobile Data Collection Without Compromising the Upper Body The G1-D Ultimate preserves everything that makes the G1-D a capable data acquisition platform — the multi-camera array, NVIDIA Jetson Orin 100 TOPS, and high-DOF upper body — and adds full mobility. Teams that need to collect manipulation data across multiple environments, workstations, or physical configurations no longer have to choose between a capable upper body and a mobile base.
Dual Battery System — Upper Body and Chassis Independent The G1-D Ultimate runs two independent power systems. The upper body quick-release 9000mAh battery provides approximately 2 hours of upper body runtime. The built-in 30Ah chassis battery provides approximately 6 hours of mobile operation. The included 51V/10A auto-charging station lets the robot return to charge autonomously — keeping it operational without constant manual intervention.
Full Chassis Perception Layer Beyond the upper body camera array, the wheeled chassis adds its own dedicated sensor stack — chassis LiDAR, dual depth cameras, a collision sensor, and a low-profile obstacle sensor. This gives the G1-D Ultimate independent spatial awareness at the chassis level, separate from and complementary to the upper body vision system.
360° Mobility at Up to 1.5 m/s Dual-wheel differential drive delivers smooth omnidirectional movement and 360° on-spot rotation at up to 1.5 m/s — giving the G1-D Ultimate the range and agility to operate across service, retail, life assistance, and industrial environments without repositioning constraints.
Five End Effector Options in One Listing Select the end effector that fits your research or application. Each option changes the total DOF count and the nature of manipulation data you can capture. All five are available as variants directly on this listing.
NVIDIA Jetson Orin — On-Board AI Processing The integrated NVIDIA Jetson Orin NX 16GB (100 TOPS) handles real-time perception, control, and on-device processing. Combined with an 8-core high-performance CPU, the G1-D Ultimate supports secondary development, custom software integration, and direct SDK access — ready for whatever data pipeline or control framework your team brings to it.
End Effector Options Select your end effector configuration at checkout. All options include dual wrist-mounted HD cameras and RGB cameras integrated into the hand unit.
Key Features of the G1-D Ultimate
Purpose-built for AI data collection — mobile platform with hardware architecture optimized for capturing embodied AI training data across dynamic environments
19 DOF core body — 7 DOF per arm × 2, 2 DOF waist, 1 DOF column lift, 2 DOF chassis; expandable to 33 DOF with dexterous hands
450mm lifting stroke — 1mm accuracy, max 60mm/s; max operating height approximately 2m
Multi-perspective camera array — head-mounted HD binocular camera (3840×1200, 115° H FOV) + dual wrist HD cameras (1920×1080, 130° H FOV)
Smart OTA updates + secondary development — open to custom SDK integration and third-party software pipelines
Visualization computer included
Handheld remote controller included
12-month warranty
A Note on Software & Data Pipelines
The G1-D Ultimate is hardware-ready for structured data collection and embodied AI workflows. Its sensor suite, compute module, and secondary development support are designed to integrate with custom or third-party data pipelines and control frameworks. RoboStore provides the hardware — we work with your team to identify the right software stack for your specific research or deployment environment. Contact us to discuss your setup.
Also Available: G1-D Standard (Fixed Base)
Need a stable, controlled environment instead of full mobility? The G1-D Standard shares the identical upper body as the Ultimate with a rigid fixed base — ideal for structured, repeatable data collection where positional consistency is the priority. [View the G1-D Standard →]
Specifications subject to change. All specs reflect manufacturer data current as of May 2026. For the most current information or custom configurations, contact sales@robostore.com.
Designed for speed and repeatability across pick-and-place, sorting, and structured manipulation tasks. With up to 120N of adjustable clamping force and a 0.24s closing time, it's the most straightforward option for high-throughput data collection.
Dex3-1: 3-Finger Dexterous Hand
Handles a wide range of grasp types such as pinch, wrap, and in-hand manipulation. With 7 DOF per hand, it's a strong choice for manipulation research that needs more than a gripper without the full complexity of a five-fingers. Tactile sensor option available.
BrainCo Revo 2: 5-Finger Dexterous Hand
The closest to human-hand dexterity on the platform. With 6 DOF across 11 joints and 0.1° repeatability, it's built for fine motor tasks and complex grasps that demand the highest fidelity manipulation data. Tactile sensor option available.
Sorting, grasping, assembling, organizing — the G1-D is designed to perform the precise, repeatable manipulation tasks that generate the high-quality embodied AI training data simulation can't replicate.
HD Binocular + Dual Wrist Cameras
A head-mounted HD binocular camera (3840×1200, 115° FOV) plus two wrist-mounted HD cameras (1920×1080, 130° FOV each) deliver synchronized, multi-perspective visual data across every manipulation task.
500mm Vertical Range. Built for Real-World Applications.
A motorized 500mm lifting column, 2-DOF waist (Z±155°, Y -2.5° to +135°), and 7 DOF per arm let the G1-D Standard operate across service, retail, life assistance, and industrial environments — at any working height, in any pose.
Advanced Mobility & Autonomous Operation.
The G1-D Ultimate adds a full wheeled chassis system — built-in battery, dedicated sensors, and auto-charging included.
Order directly through our website or use the form below to connect with a product expert first. Delivery timelines depend on model availability, but we ship as quickly as possible — and we’ll keep you updated every step of the way.
Setup & Configuration
Every robot is different. Depending on your selected model and accessories, configuration can be plug-and-play or require a tailored setup. We offer optional demos, consultations, and ongoing support to ensure you’re up and running with confidence.
Get up and GO
Once set up, your robot is ready to roll. Preloaded actions make it easy to get started, while manual controls and custom routines let you push the limits of what’s possible.
Choosing a selection results in a full page refresh.
