Proof today4-pod prototype
Four pods streaming pressure, contact location, and temperature. Demonstrated live at Cal Hacks, October 2026.
Modular tactile pods for existing grippers
Dermis is a modular tactile sensing kit for existing robot grippers and hands. Place sensing pods where contact matters, replace individual pods when they wear out, and verify each replacement before the robot goes back to work.
4-pod prototype live · pressure, contact location and temperature streaming today
01The problem
Teams have to choose sensors, design electronics, mount them, calibrate them and maintain them. When one sensing element wears out or fails, replacement can mean more integration and recalibration work.
On tasks like key removal and plug insertion, policies that use touch have been shown to outperform vision-only setups. That is why teams want sensing on grippers — and why the integration and maintenance cost matters.
Examples from published contact-heavy tasks. Full citations are in Research and sources.
02Idea
Dermis turns tactile sensing into a replaceable component instead of a custom sensing stack. We are not building the robot hand. We are building the touch layer teams add to the grippers they already have — and swap one location at a time when a pod wears out.
Four pods streaming pressure, contact location, and temperature. Demonstrated live at Cal Hacks, October 2026.
The loop
Attach. Verify. Resume.
Attach pods to an existing gripper, verify them, then sense. When one location wears out, replace that pod, verify it, and resume.
Starting with existing robot grippers and hands used in contact-heavy work: gripping, insertion, and delicate handling.
What makes it different
The hassle is not finding a sensor. It is deploying and maintaining one.
After replacing a pod, run a defined calibration and verification check before returning the robot to operation. Dermis is being designed so that check is repeatable, not a one-off lab procedure.
If one sensing location fails, replace that pod instead of replacing or rebuilding the entire sensing assembly.
One sensor architecture, task-specific mounts. We are starting with selected grippers and expanding compatibility as we validate deployments.
Pods, a hub, task-specific mounts, and software for recording and verifying tactile data — so sensing is not another custom stack the team has to own.
| Who | Existing-hardware integration | Individual replaceability | Post-replacement verification | Modular mounting |
|---|---|---|---|---|
| In-house custom stackssensors, PCBs, placement, firmware | You design it | Often rebuild | Left to each lab | Custom each time |
| Robot-side tactile sensorse.g. XELA uSkin, Contactile | Yes | Vendor process | Often a paid add-on | Vendor mounts |
| Open magnetic skinse.g. AnySkin, OSMO | Build it yourself | Skin can be swapped | Left to each lab | Left to each lab |
| DermisReplaceable tactile pods | Current focus | Prototype architecture | In development | Task-specific mounts |
Why it matters
Touch should be something you attach, check, and keep using — not another system the team has to maintain.
How teams use Dermis
attach ⇄ resume
Teams add touch to the hardware they already run, instead of choosing parts, laying out boards, and owning the stack.
A failed patch is a part swap and a check, not a rebuild of the sensing assembly.
A worn or failed location is a pod swap plus a verification check, not another integration project.
We are also testing the same sensing architecture for human demonstration collection. That work is experimental and not the v1 product.
03How it works
Snap pods onto an existing gripper where contact happens.
Run a check before the robot works, so the setup is verified, not assumed.
The current prototype streams pressure, contact location, and temperature.
If one location wears out or fails, swap that pod, not the whole assembly.
Run the same calibration and verification check, then return the robot to operation.
Our current four-pod prototype streams tactile data live. We are now productizing the mounting, replacement and calibration workflow for robot grippers.
04Product
Existing gripper, Dermis pods, a replacement workflow, and verification. Starting with robotics teams working on contact-heavy manipulation such as gripping, insertion and delicate handling.
Small, swappable tactile patches. Magnetic silicone skin over a magnetometer and an IMU on each pod.
The v1 product is an existing gripper plus Dermis pods, with a replacement and verification workflow.
An ESP32 hub reads the pods over SPI today. Firmware is C/C++ on ESP-IDF.
Python for calibration and processing, live streaming over WebSocket, recordings in JSON.
Illustrative traces. The prototype streams pressure, contact location, and temperature. Mounting, replacement, and calibration for grippers are what we are productizing now.