Skip to content
Droid/Robot hand evidence record

TESOLLO DG-5F-M

Separate the direct ROS control path from the x86-64 SDK bridge, and specify optional fingertip sensing and wrist configuration before ordering.

20-DoF model; distinguish standard and Short Wrist

Does a ROS 2 label mean every DG-5F-M software path runs on a Jetson?

No. TESOLLO documents a precompiled SDK bridge for x86-64 only and says to exclude it on ARM64. The direct ros2_control path uses separate hardware and TCP packages. That offers a different integration route, but this review does not establish a tested Jetson configuration.

TESOLLO ↗ TESOLLO ↗

Hardware and interface evidence

Original units and conditions are retained. A published interface does not establish a completed or independently tested integration.

Model-specific claims and evidence limits.
FieldReported valueConditions and source
Actuation20 independently actuated jointsFour per finger, with one integrated actuator per joint.
TESOLLO ↗
Input supply24 VNominal specification; confirm allowable tolerance, wiring and transients in the delivered manual.
TESOLLO ↗
Maximum current consumption10 AManufacturer maximum, not typical operating current or average energy consumption.
TESOLLO ↗
Reported mass1,763 gProduct-page figure; selected wrist and sensor options must be confirmed.
TESOLLO ↗
CommunicationModbus RTU/TCP; Ethernet TCP/IPThe ROS hardware driver requires Developer Mode and its custom Ethernet protocol, not the Operator Mode Modbus path.
TESOLLO ↗
Reported control frequency250 HzProduct control-frequency claim, not measured glove-to-actuator latency.
TESOLLO ↗
Fingertip options6-axis F/T, 3-axis force or tactileOptional configurations; do not assume these sensors are included in the quoted base hand.
TESOLLO ↗
ROS 2 softwareHumble and Jazzy listed; Lyrical experimentalThe repository says Lyrical is not validated on hardware. Pin model and common-package submodules.
TESOLLO ↗

What this means for the control loop

Droid’s analysis of the cited documentation:

  1. Choose the software path before buying the host. Excluding the x86-only bridge removes one documented binary dependency; it does not prove that the remaining controller is validated on your ARM64 board.
  2. Specify the fingertip sensor and wrist as line items. Optional sensing changes the data available for contact feedback, and the Short Wrist variant does not inherit every peripheral feature.
  3. The manufacturer's MANUS example maps ergonomic glove values into joint commands. Its rule-based scaling needs operator and task calibration; it is not evidence of autonomous grasp success.

TESOLLO ↗ TESOLLO ↗ TESOLLO ↗

Documented components and software paths

Included sensing, optional hardware and software dependencies are different relationship types. These records do not infer undisclosed suppliers or part numbers.

documented option

TESOLLO fingertip F/T option

Optional fingertip configuration, not a claim about the shipped bill of materials.

6-axis option; part number unspecified

TESOLLO ↗
documented option

TESOLLO fingertip tactile option

Confirm the model-specific tactile hardware and software interface.

Model-specific configuration to confirm

TESOLLO ↗
documented software path

TESOLLO SDK-to-ROS bridge

Optional SDK bridge uses an x86-64 library. The direct ROS hardware path is distinct.

Workspace 1307f7d; x86-64 precompiled library

TESOLLO ↗

MANUS ROS stream → DG-5F-M

TESOLLO publishes a ROS 2 retargeting node that maps MANUS ergonomic data into DG5F joint references. The model-specific package is identified as DG-5F-M by its parent workspace.

The example does not identify an exact MANUS glove or firmware combination. It is a rule-based mapping with calibration factors, not evidence of a tested production deployment or autonomous task success.

Inspect the retargeting evidence →

Still to establish

  • Exact hardware, wrist, fingertip and firmware configuration
  • Validated ARM64 host and controller build
  • Mounting pattern and arm payload margin for the complete tool
  • Delivered price and measured task performance

Before enabling the full workflow

  1. Match the physical hand and wrist variant to the model files.
  2. Verify Developer Mode and Ethernet for the direct ROS driver.
  3. Confirm optional sensors are fitted and exposed by the selected API.
  4. Measure timing, contact behavior and recovery with the real task.

Source record

  1. TESOLLO ROS 2 workspace and architecture limits ↗TESOLLO · Commit 1307f7d; pinned per-model submodules · Checked 2026-11-10
  2. DG-5F-M model packages ↗TESOLLO · Submodule commit 9e9af12 selected by workspace 1307f7d · Checked 2026-11-10
  3. ROS 2 connection and device-mode prerequisites ↗TESOLLO · Developer Mode + Ethernet path; model-specific panel differences · Checked 2026-11-10
  4. MANUS stream to DG5F retargeting ↗TESOLLO · Commit 9e9af12; glove model and firmware unspecified · Checked 2026-11-10
  5. DG-5F-M product specification ↗TESOLLO · DG-5F-M; Short Wrist is a separate option · Checked 2026-11-10
START WITH THE WORKFLOW

Bring us the task.
Let’s find the fit.

Share your environment, the job to be done, and the outcome you need to measure.

Request a project review