New Product Launch: ExoArm-7 Pro Brings Natural Dual-Arm Teleoperation to Embodied AI Research

Developing capable robots requires more than designing stronger hardware or larger AI models. Researchers also need a practical way to transfer natural human movements to robotic systems and collect high-quality demonstration data for imitation learning, embodied AI, and Vision-Language-Action model development.

To support these workflows, Foxtech Robotics is introducing the ExoArm-7 Pro Wearable Exoskeleton, a lightweight dual-arm motion-capture and teleoperation system designed for robotics research, human demonstration data collection, and VLA dataset creation.

Combining 14-degree-of-freedom arm tracking, high-precision encoder feedback, high-speed sampling, dual-IMU sensing, cross-platform software, and an official SDK, ExoArm-7 Pro gives research teams a more direct way to capture human upper-limb motion and map it to robotic platforms.

For projects that also require detailed finger and hand data, the system can be expanded with the optional ExoGlove Pro teleoperation data gloves.

Capture Natural Dual-Arm Motion with 14DOF Tracking

Many manipulation tasks involve coordinated movement across both arms rather than a single end-effector. Picking up a large object, opening a container, organizing tools, folding materials, or transferring an item between hands all require synchronized motion and continuous changes in joint position.

ExoArm-7 Pro uses 14 passive joints, with seven joints on each arm, to capture these natural human movements. Instead of asking an operator to control each robot joint independently, the wearable structure records the operator’s own arm motion and converts it into data that can be used for robot teleoperation or human demonstration collection.

This approach makes it easier to collect coordinated bimanual trajectories while allowing the operator to perform tasks in a familiar and intuitive way.

High-Precision Feedback for More Accurate Motion Mapping

Useful teleoperation depends on more than detecting large arm movements. Small changes in joint angle can affect the position and orientation of a robotic arm, particularly during close-range manipulation or coordinated dual-arm tasks.

ExoArm-7 Pro provides 14-bit joint data for fine-grained position feedback. Its in-house-developed encoder joint modules are designed to detect subtle angular changes, helping improve the precision and stability of human-to-robot motion mapping.

Each joint module weighs less than 70 g, supports up to 10 N·m of load-bearing torque, and includes an adjustable mechanical limit. This modular design helps balance motion-capture performance with the need for a lightweight wearable structure.

Up to 1000Hz Sampling with Millisecond-Level Low Latency

When a robot follows a human operator, delayed or low-frequency data can make movement feel disconnected and may reduce the quality of the collected demonstration.

ExoArm-7 Pro supports joint-data sampling rates of up to 1000 Hz and millisecond-level low-latency control. The high update rate enables the system to capture fast motion changes and provide responsive data for real-time teleoperation.

For researchers building imitation-learning or VLA datasets, this also provides a dense motion record that can better represent how a task develops over time, from the initial approach to manipulation and completion.

Dual-IMU Sensing for More Complete Motion Information

Joint angles provide essential arm-motion data, but orientation and broader body movement can also affect teleoperation. ExoArm-7 Pro therefore includes dual-IMU sensing to expand motion and orientation tracking.

The standard package includes a back-mounted IMU sensor and an external IMU sensor. Together with joint data, these sensors provide a more complete basis for interpreting operator movement and integrating the wearable system with different robotic platforms.

Lightweight Design for Longer Data-Collection Sessions

Wearable systems must remain practical during repeated demonstrations. Excessive weight can increase operator fatigue, restrict natural motion, and reduce consistency across long data-collection sessions.

ExoArm-7 Pro integrates its dual-arm structure into a 2.6 kg ergonomic wearable design. Its lightweight construction is intended to make operation more comfortable while maintaining the precision required for robotics development.

The system also offers custom control handles and an optional glove adapter, allowing teams to configure the operator interface according to different robots and manipulation tasks.

Extend Arm Tracking to Detailed Hand Capture with ExoGlove Pro

Arm motion alone is sufficient for some teleoperation tasks, but dexterous manipulation requires additional information about the hands and fingers. Grasping small objects, operating tools, rotating components, or manipulating flexible materials can depend on detailed finger posture and fingertip movement.

The optional ExoGlove Pro extends ExoArm-7 Pro from dual-arm tracking to hand and finger motion capture. Using five proprietary VetraSense sensors, the glove system can reconstruct a complete 25DOF hand pose. Each fingertip supports 6DOF tracking, while adjustable vibration feedback can provide an additional interaction channel for teleoperation.

ExoGlove Pro supports data output at up to 180 Hz and wired latency below 10 ms. Multi-platform SDK support and motion-retargeting algorithms help developers map captured hand movements to different dexterous hands and robotic platforms.

ExoGlove Pro is an optional accessory sold separately and is not included in the standard ExoArm-7 Pro package.

Built for Integration Across Robotics Platforms

Research teams often work with different robot arms, dexterous hands, operating systems, and control architectures. A data-collection device is only useful if it can be connected to the rest of the development environment.

ExoArm-7 Pro provides an official SDK and free integration support to assist with robotic platform development. Its visual software interface supports Windows, macOS, and Linux, giving laboratories, universities, and robotics companies greater flexibility when building their teleoperation and data pipelines.

The combination of configurable handles, optional glove integration, high-frequency joint data, and cross-platform software makes the system suitable for both early-stage research prototypes and more structured data-collection projects.

Applications for Embodied AI and Robot Learning

ExoArm-7 Pro is designed for a range of research and development workflows, including:

• Dual-arm robot teleoperation
• Embodied AI data collection
• Human demonstration recording
• Imitation-learning dataset creation
• VLA data collection
• Dexterous manipulation research
• Robotics education and laboratory development

Researchers can use the system to demonstrate manipulation tasks directly, collect synchronized motion trajectories, and build datasets that reflect how humans coordinate both arms during real operations.

What Is Included?

The standard ExoArm-7 Pro package includes:

• 1 × ExoArm-7 Pro Wearable Exoskeleton
• 1 × Back-Mounted IMU Sensor
• 1 × External IMU Sensor

The optional ExoGlove Pro package includes:

• 2 × Motion Capture Gloves
• 1 × Wireless Receiver and Battery Module
• 2 × USB-C Data Cables

A More Direct Connection Between Human Motion and Robot Learning

The quality of robot behavior depends heavily on the quality of the data used to train and control it. By capturing natural dual-arm movements at high speed and extending the workflow to detailed hand motion when required, ExoArm-7 Pro helps bridge the gap between human demonstration and robotic execution.

For teams developing embodied AI, VLA models, dual-arm manipulation, or dexterous teleoperation, the system provides a wearable and integration-ready foundation for turning human actions into structured robot-learning data.