UNITREE

End-to-End Platform for Humanoid Robot Data & Training

IHigher-DOF Robot Platform

Total Degrees of Freedom

Robot DOF(Excl.End-Effector):19
Arm Degrees of Freedom:7x2
Waist Degrees of Freedom:2
Column Degrees of Freedom:1
Base Degrees of Freedom:2

Expanded Operational Workspace

Mobile Operation :

Adopting a mobile lifting design that combines wheels and lifting mechanisms

VerticalWorkspace:0-2m Waist ROM (Z):±155° Waist R0M(Y):-2.5°\~+135°

ILower-Latency Control Response

Control Accuracy :

·Lifting Accuracy: ± 0 . 5 \mathsf {mm }
·End-Effector Gripper Accuracy: ± 0 . 1 \mathsf {mm } （20
\*Note:Accuracyvaries withdiferent end-effector configurations.

Latency & Dynamic Response :

·System Teleoperation Latency: <100ms ·Sampling Rate: 60 Hz

Streamlined Data Acquisition Tools

Streamline data collection and reduce costs with flexible,standardized workflows 一moving beyond traditional manual processes.

Key Features

Visual Template Management for Efficient Collection

Integrating project management, task allocation,progress tracking,and statusanalysis intoa unified platform.With templated configurations,data collection tasks can be generated with one click.Real-time tracking of the entire workflow ensures smoother collaboration and more efficient dataacquisition.

Flexible Configurations with Diverse Platforms& End Effectors

The system supports data collectionacross various robot platformsand end-effector configurations.Its robust data standardization capability enables end-to-end processing from diverse devices to high-quality model training data.

High-Concurrency Architecture for Scalable Operations

Equipped with the technical capability to support hundreds of robots performing synchronous data collection.Through high-concurrency architecture and load-balanced scheduling, the platform ensures real-time reception and processing of massive data streams.

24/7 Online Collection, StableandReliable

Builtona highlyavailableservicearchitecture,theplatform enables reliable 24/7 data collection.With extensive format compatibility,collected data can be directly output or converted into mainstream training formats,accelerating R&D efficiency.

Data Acquisition Pipeline

Comprehensive Model Training & Inference Tools

It supports an end-to-end workflow from data processing to one-click model deployment.The platform seamlessly integrates a variety of mainstream open-source robotic model frameworks.

Key Features

Rich Ecosystem with Mainstream Model Support

Anopen model ecosystem is established,featuring built-in communitydatasetsand support foropen-source dataset training.It offers deep integration with leading open-source models such as Pland GROOT.

Simulation Environment for Model Evaluation

02

Includesa high-fidelity, high-precision3Dasset library. By constructingrealistic simulation environments,itrapidly generates comprehensive evaluationschemesto support algorithm validation.

User-Friendly Design for RapidDeployment

Fully deployable out-of-the-box,minimizing setup time. Launch model development instantly via"one-click training." leverage integrated simulation tools forreliableevaluation,and achieve seamlessmigration fromalgorithmsto real-world machines.

Distributed Training for High Efficiency

04

Builton a high-performance distributed training architecture, the platform enables elastic schedulingof computing tasks and parallel acceleration.It dynamically scales based on workload resources,achieving up to 90% GPU utilization.

Training and Inference Pipeline

UNITREE

Web:www.unitree.com Tel:+86 18767138485 Email:sales_global@unitree.com

×□ Follow us@Unitree Robotics

Please visit Unitree Robotics Website for more related terms and policies, andcomplywithlocallawsandregulations.