Fields & coordinates
Confirm handedness, joint names, units, orientation representation, reference frames and missing-data rules. Outputs depend on the SDK and firmware version.
FORMOTION / FINGERTIP TRACKING / EMF + IMU
G7 magnetic-inertial data gloves
Fingertip tracking for dexterous robot teleoperation.
Capture hand motion during pinching, grasping and finger opposition. Manufactured by Forsense; TOBI supports selection, Ego camera combinations and delivery coordination. Robot integration is assessed per project.
120 HzSensor sampling
2 mmPosition accuracy · specified
<20 msWired end-to-end latency
Source: Forsense datasheet · confirm configuration and test conditions
FOR MOTION. FOR DATA.G7
Human motion → hand data → project integration
01 / FIT FOR YOUR TASK
Start with the task.
Choose the right input.
Where it fits
Tasks requiring fingertip spatial positions, with a controllable magnetic environment and a validated connection topology.
What to verify
The specified 2 mm position accuracy is not robot end-effector accuracy. Evaluate nearby metal and close-hand operation. Camera synchronization and robot retargeting are not included by default.
Explore the alternative · G2-L →02 / CAPTURE TO CONTROL
From human movement
to a robot task.
The glove provides the input. Mapping, control and synchronization define the complete workflow. This is an integration diagram, not a claim of validated robot compatibility.
- 01 / CAPTURE
Capture
Record human-hand motion
- 02 / SDK
Decode
Confirm fields, frames and timestamps
- 03 / PROJECT
Retarget
Map to the target joints and limits
- 04 / PROJECT
Control & record
Connect the controller; log states and outcomes
Dashed stages require project validation. The robot hand, controller and cross-device synchronization services are not included by default.
03 / DATA & INTEGRATION
Define the data.
Then validate integration.
Define interfaces, outputs and acceptance criteria before engineering integration.
Timing & quality
Specify device and host timestamps, latency definitions, jitter, dropped frames and drift. Retain processing records for resampled data.
Target robot
Confirm degrees of freedom, joint order, joint limits and the communication protocol. Scope follows validation on the actual robot model.
Public technical resources
Manufacturer specifications and integration resources are open. A validated TOBI recording sample and robot compatibility list are not published here; request a technical assessment if needed.
| Item | Current evidence | Before integration |
|---|---|---|
| Device output | Manufacturer specifications and SDK documentation | Firmware, SDK, field definitions and operating system |
| Robot-hand control | Project validation required | Target hand, driver, retargeting and task acceptance |
| ROS 2 / LeRobot / Isaac Lab | No validated integration version published here | Target framework version, conversion code and test sample |
04 / CHOOSE YOUR GLOVE
G7 or G2-L?
Choose by capture task.
| Criterion | G7 | G2-L |
|---|---|---|
| Capture focus | Fingertip spatial position and hand motion | Gestures, articulation and hand orientation |
| Tracking | EMF + IMU magnetic-inertial | IMU inertial |
| Rate definition | 120 Hz sensor sampling | 50 Hz data transmission |
| Latency definition | End-to-end: wired <20 ms / wireless <30 ms | Signal latency ≤20 ms |
| Connection and runtime | Wired Hub; optional wireless >3.5 h | Bluetooth 5.1; 7 h battery life |
| Weight per glove | 115±5 g | 90±5 g |
| Before choosing | Magnetic environment, hand separation, wiring and position reference frame | Calibration, drift, Bluetooth coverage and pose-data requirements |
05 / SPECIFICATIONS
Specifications,
in context.
The following are manufacturer specifications, not independent TOBI measurements. Performance depends on configuration, environment, calibration and test method.
Open the source datasheet ↗Sources checked: 2026-10-01
| Sensor architecture | 5 fingertip EMF+IMU modules plus a back-of-hand IMU per hand |
|---|---|
| Sensor sampling rate | 120 Hz |
| End-to-end latency | Wired <20 ms; wireless <30 ms |
| Spatial accuracy | Position 2 mm; orientation 3° (manufacturer specification) |
| IMU | 6DOF; gyroscope ±4000 dps; accelerometer ±8 g |
| Wired communication | CANFD or UART; UART can convert to 485 / 232 / 422 / USB |
| Power | Hub USB-C; 4.6–5.5 V, approximately 300 mA |
| Optional wireless module | Approximately 1.7 W; >3.5 h runtime; approximately 10 m range; module approximately 65 g |
| Fit and weight | 115±5 g per glove; M / L / XL, standard L |
| Environment | 0–40 °C; humidity 20%–80% |
| Module dimensions | Fingertip 12.6×22.85×7.7 mm; hand controller 67.4×53.4×22.2 mm |
06 / CONFIGURATION
Know the scope
before procurement.
Hardware configuration
The datasheet lists a glove pair, hand controllers, G7-Hub, cables, power adapter and storage accessories.
Options & peripherals
Wireless modules, forearm modules, TTL-to-USB-C and wireless receiver; quantities and versions require a confirmed bill of materials.
Project services
Robot integration, camera synchronization, schema conversion and acceptance support are assessed per project. Pricing, lead time and services are itemized in the quotation.
07 / QUESTIONS, ANSWERED
Before you choose.
Understand capabilities and limits before committing to an integration.
Is this a robot hand or a data glove?
This is a wearable human-hand motion capture input device. A robot hand is the actuator; SDK integration, retargeting and a robot control interface connect the two.
Which robots can it control directly?
No model-specific robot validation list is published here. Provide the target hand, degrees of freedom, interfaces and task to assess development and testing. Plug-and-play compatibility is not assumed.
How should I choose between G7 and G2-L?
Evaluate G7 for fingertip spatial position, including the magnetic environment. Evaluate G2-L for wireless hand-pose capture. Rate and latency definitions differ; validate against your task.
Does it provide tactile sensing or force feedback?
The specifications cited here do not establish tactile sensing, contact-force capture or force feedback. Confirm hardware and peripherals separately if required.
How does synchronization with E6 / E2 work?
Confirm both clocks, timestamps, alignment method and error criteria. G7 Hub synchronization between hands is not hardware synchronization with an Ego camera. Starting simultaneously is not time alignment.
Is the data ready for LeRobot or VLA training?
Hand trajectories can be part of demonstrations but are not automatically robot actions. Validate the action space, robot states, aligned vision, schema and target framework version before conversion and training.
How do I get a quotation, lead time or evaluation?
Review the public specifications, then send the model, application and email. Pricing depends on configuration, accessories, quantity and services. Confirm lead time, evaluation availability and warranty in writing.
PROJECT ASSESSMENT
Tell us about your task.
Define the right setup.
Expect a reply within one working day to discuss the model, interfaces, sizing and delivery scope. Basic specifications are available without a form.
+86 137 5502 0164For robot integration or camera synchronization, include the target equipment and task.