Dexterous Hand_Embodied AI-SHANGHAI CHINGMU VISION TECHNOLOGY CO., LTD

Business pain points

Frequent finger occlusion causes data interruption and loss

Fingertip-level micro-motions (e.g., pinch grasp) difficult to identify and quantify

Single-sensor limitations: optical struggles with occlusion, IMU suffers drift

High teleoperation latency leads to delayed, unresponsive robot control

Variable experimental setups with complex integration and high adaptation costs

Our advantages

Patented tracking algorithm with robust occlusion resistance—no loss even in complex gestures

Sub-millimeter capture precision with synchronized multimodal data (motion, haptics, etc.)

Dual-modal fusion algorithm, precise and comprehensive

Millisecond-level latency ensures real-time motion consistency

Flexible integration with one-click finger motion assessment reports

Introduction

In response to the core challenges of training data acquisition and performance evaluation for robotic hands, CHINGMU offers a comprehensive solution. Our system provides real-time 6DoF pose tracking for both human and robotic hands, enabling low-latency, high-fidelity mapping for imitation learning and teleoperation while ensuring stable data output even under severe occlusion.

Applications ：R&D validation, Fine Manipulation, HRI, Teleoperation，Imitation Learning

Data acquisition

SYSTEM COMPONENTS

Flagship Camera: K Series

High-end optical mocap cameras built for demanding precision tracking.

Reference Camera:R Series

High-performance RGB reference camera for synchronized video, timecode alignment, and motion data validation.

Other Camera: MC Series

Flexible optical mocap cameras for large-space, high-precision tracking across diverse production setups.

PULSEH Optical-Inertial Fusion Glove

Captures precise multimodal finger motion data with sub-millimeter accuracy and robust tracking stability.

Software

CMAvatar enables real-time 6DoF tracking, motion visualization, data synchronization, and quantitative analysis for robot training and evaluation.

Advantages of the plan

Multimodal Data AcquisitionIntegrates optical, inertial, and tactile sensing to synchronize motion, video, tactile, object, and environmental data. This enables high-quality hand-object interaction datasets for imitation learning, dexterous manipulation research, and embodied AI training.

Consult >

Robust Anti-Occlusion TrackingAdvanced tracking algorithms maintain stable finger pose estimation during complex gestures and severe visual occlusion, minimizing data dropouts and ensuring reliable data flow for teleoperation and AI training.

Consult >

High-Precision Finger CaptureSupports both optical and opto-inertial modes to balance ease of use, motion detail, and tracking stability. The system captures subtle fingertip movements such as pinch, grasp, rotation, and fine manipulation with sub-millimeter-level precision.

Consult >

Low-Latency Feedback & ControlFull-link optimization enables millisecond-level data output and seamless integration with teleoperation pipelines, significantly improving motion responsiveness, control fluency, and operator-to-robot consistency.

Consult >

Quantifiable Performance MetricsBuilds a multidimensional metric system and automatically analyzes key indicators, including motion accuracy, positional repeatability error, trajectory deviation, pose stability, and operational consistency. Supports custom eva1uation dimensions and pass/fail thresholds for different R&D stages.

Consult >

System Flexibility & ScalabilityFlexible camera and marker configurations support different laboratory environments, robot platforms, and research workflows. The system can integrate with external sensors and simulation tools to complete the teach-train-validate R&D loop.

Consult >

Related Cases

CHINGMU Dynamic Real-time delivery

Qingtong Vision × Xiying Group: A New Chapter in Digitalization of Film and Television / Company Updates

In the field of film art, when the movements of characters are smooth and fluid, the expressions of digital characters are vivid and expressive, and the virtual shooting seamlessly integrates with the real scenes, the success of such works is inseparable from the strong support of motion capture technology.

2025-10-06

Qing Tong Vision × UITM: Overseas Empowerment of Sports Science / Company Updates

Under the global trend of upgrading global sports research towards more refined and quantitative analysis, universities and research institutions around the world are accelerating their exploration of the integration path between technology and academic research.

2025-10-06

A virtual reference trajectory scheme for tracking control of wheeled mobile robots with slip disturbances / Engineering

2026-05-19

IROS 2025 Zhejiang University: Research on Dextrous Manipulation of Dual Robots / Industry News

The dexterous operation of dual robots, due to its high degree of freedom coordination, multimodal perception fusion, and complex skill combination, has become an important research direction in the field of robotics.

2025-10-06

IROS 2025：Blue Eye Motion Capture Black Technology / Company Updates

As the annual event IROS 2025 of the robotics circle approaches, as the "prominent representative" of motion capture technology in the industry, CHINGMU Qing Tong Vision has already prepared a "full-strength surprise": multiple hardcore workshops, in-depth visits to industrial parks,

2025-10-06

Obtain exclusive case studies and industry solutions

Get the complete case in one click. Our experts will precisely match the most suitable motion capture solution for your industry, helping you reduce costs and increase efficiency.

Shanghai Jiao Tong University | Optimizing Dexterous Hand Control with High-Precision Motion Data

Zhejiang University | Optical Motion Capture-Based Teleoperation Training System

Tsinghua University | Advancing Dexterous Hand Research with CHINGMU Motion Capture

Obtain exclusive case studies and industry solutions

Name*

Tel*

Email*

Enterprise / Institution*

Country*

Address*

Application fields*

Please leave a message.*