Description
- Unitree G1-D Standard B Humanoid Robot Platform (Dex 3-1 Hands)
- Multi-sensor perception system
- Secondary development supported; visualization computer included
- Fixed-base dual-arm humanoid robot
- Dex3-1 force-controlled three-finger dexterous hands (w/ RGB Camera & wo/ tactile feedback)
- Total degrees of freedom: 31
The Unitree G1-D Standard B Humanoid Robot Platform (Dex 3-1 Hands) is the fixed-base G1-D Standard B configuration that upgrades the end effectors to Dex3-1 force-controlled three-finger dexterous hands (no tactile feedback, with RGB camera), designed for more articulated, finger-level manipulation than the two-finger grippers.
The G1-D platform is built around a full-size humanoid upper body with a 17-DOF kinematic structure (excluding the end effector), a vertical lifting column with a 500 mm stroke for adjustable working height, and an integrated multi-sensor perception stack (head-mounted binocular vision plus wrist cameras) to support manipulation and teleoperation workflows; it is supplied with the standard control electronics/software stack and the required power/charging interface for deployment. As an authorized reseller with local support and logistics, RobotShop acts as a single point of contact for purchase and after-sales service, reducing friction versus a direct-from-manufacturer purchase and avoiding the need to ship the robot abroad for support or resolution.
Genuine products with manufacturer-approved warranty coverage.
Customs, inspection, and prep handled locally before shipment.
Sales, support, and returns managed by one team.
Docs, tutorials, and community answers to move faster.
for Humanoid Robot
Proprietary humanoid platforms built end-to-end with in-house actuators, gearsets, encoders, and integrated sensing.
A single workflow covering the entire data lifecycle: capture, processing, annotation, validation, and dataset asset management.
Supports distributed training, custom model creation, and smooth deployment, with compatibility across leading open-source frameworks.
Accelerate data collection and lower operating costs with flexible, standardized workflows — replacing slow, manual methods with a repeatable process.
Combines project coordination, task assignment, progress monitoring, and status reporting into one workspace. Using reusable templates, collection tasks can be created in a single click. Real-time visibility across the pipeline improves collaboration and keeps data acquisition running smoothly.
Supports data capture across multiple robot platforms and end-effector configurations. Strong standardization turns inputs from diverse devices into consistent, high-quality training-ready datasets through an end-to-end workflow.
Designed to coordinate hundreds of robots performing synchronized data collection. With high-concurrency services and load-balanced scheduling, the platform reliably ingests and processes large, real-time data streams.
Built on a highly available service foundation to support dependable 24/7 collection. Broad format compatibility lets outputs be used directly or converted into widely adopted training formats, speeding up R&D cycles.
-
1Create New
Data Collection
Tasks -
2Task Updates
and
Assignment -
3Data Capture
and
Annotation -
4Upload Data
and Review -
5Data Archiving
-
6Dataset Export
Covers the complete workflow from data preparation to one-click deployment. The platform integrates smoothly with widely used open-source robotics model frameworks.
An open model ecosystem with built-in community datasets and support for training on open-source datasets. It also provides deep integration with leading open-source models such as PI and GROOT.
Includes a high-fidelity, high-precision 3D asset library. By building realistic simulation scenes, it quickly produces comprehensive evaluation plans to support algorithm verification.
Ready to deploy immediately, reducing setup time. Kick off development with "one-click training," use integrated simulation for reliable evaluation, and transition smoothly from algorithms to real-world machines.
Built on a high-performance distributed training stack, enabling elastic task scheduling and parallel acceleration. It scales dynamically with available resources, achieving up to 90% GPU utilization.
-
1Choose
Model
Architecture -
2Set Training
Configuration -
3Live Training
Monitoring -
4Parameter
Tuning -
5Run Simulation
Tests -
6Export & Deploy
Model
Application Scenario
Service
Life
Retail
Industry
An Open World-Model-Action (WMA) Framework
Action Generation Guided by
Accurate Prediction
High-Fidelity Feedback for Data
Synthesis
Included
| • | G1-D-U2 robot (fixed-base dual-arm) |
| • | Smart battery (quick-release) 9000mAh |
| • | Main unit charger 54V 5A |
| • | Handheld remote controller |
| • | Dex3-1 three-finger force-controlled dexterous hands (no tactile; includes RGB camera) x2 |
Package contents vary by configuration.
| • | 1260 × 500 × 500 mm |
| • | 1680 × 500 × 500 mm |
| • | Approx. 50kg |
Values shown are the overall dimensions at minimum and maximum column height.
| • | Dex3-1 three-finger dexterous hands (without tactile feedback) |
| • | Regular caster wheel base |
| • | Jetson Orin NX 16GB (100TOPS) |
| • | Head HD Binocular (*1) & Wrist HD Camera (*2) |
[1] The maximum load of the arm varies greatly under different arm extension postures.
[2] For end-effector selection, please contact our sales team.
[3] For more information, please read the secondary development manual.