Paper

OpenTac: An Open Ecosystem for Customizable, Scalable, and Transferable 3D Tactile Sensing

Ariel Slepyan^{J, M}, Geeta Chandra Raju Bethala^J, Jiafeng Gu^J, Anagha Rajendra Dangle^J, Yinzhe Zhou^J, Hongyuan Fu^J, Jade Sim^J, Tuhin Basu^J, Xinyi Yin^J, Md Faizal Karim^J, Gregory Hager^J†, Homanga Bharadhwaj^J†, Anand Bhattad^J†, Krishna Murthy Jatavallabhula^J†, Nitish V. Thakor^J†

^J Johns Hopkins University  ·  ^M Massachusetts Institute of Technology  ·  ^† Equal advising

Contact: Ariel Slepyan, aslepyan@mit.edu; Nitish V. Thakor, nitish@jhu.edu

Paper Video Code

Abstract

We present OpenTac, a novel sensor and a full-stack ecosystem for tactile skins that are highly customizable, scalable, and readily deployed in downstream robot control and learning applications. Our core contribution is twofold. First, we develop a scalable resistive tactile skin with a structured deformable overlay that converts tangential loading into spatially varying pressure. Differential responses across neighboring ‘taxels’ support decoding of local shear vectors, and the common-mode response provides normal-force information. Second, we provide a highly customizable sensor design tool that is able to automatically generate fabrication files for these tactile skins from 2D outlines and developable 3D surfaces, including from abstract descriptions such as text (e.g., “design a sensor for a size 9 shoe insole”). We develop an ecosystem around the OpenTac sensor, including the CAD design tool, a sensor simulator, cross-sensor transfer, and fast/adaptive data acquisition. We demonstrate the broad applicability of the OpenTac-CAD sensor design tool by designing and fabricating more than 30 sensors across a variety of robot embodiments. We characterize the three-dimensional tactile fields produced by OpenTac with a commercial, high-precision force/torque sensor. We demonstrate the broad applicability of 3D tactile sensing across contact-rich feedback control and manipulation policy learning on four real-world robot embodiments. Additionally, adaptive acquisition enables kilohertz-rate partial readouts, exceeding full-array raster scanning on the same sensor by 19.2× at 16 sampled taxels per update on a 1024-taxel skin.

The OpenTac ecosystem. (A) Custom tactile skins for normal and shear sensing on robot hands, grippers, arms and feet. (B) OpenTac-CAD compiles a text description, 2D outline or 3D surface into a flex-PCB tactile sensor design, including the sensing layout, routing and shear-bump case. A shared sensor manifest connects the design to OpenTac-Sim, OpenTac-Transfer and OpenTac-Xampling. (C) Deployments on four robot platforms.

Citation

@misc{opentac2026,
  title         = {{OpenTac}: An Open Ecosystem for Customizable, Scalable, and Transferable {3D} Tactile Sensing},
  author        = {Slepyan, Ariel and Bethala, {Geeta Chandra Raju} and Gu, Jiafeng and Dangle, {Anagha Rajendra} and Zhou, Yinzhe and Fu, Hongyuan and Sim, Jade and Basu, Tuhin and Yin, Xinyi and Karim, {Md Faizal} and Hager, Gregory and Bharadhwaj, Homanga and Bhattad, Anand and Jatavallabhula, {Krishna Murthy} and Thakor, {Nitish V.}},
  year          = {2026},
  url           = {https://opentac.github.io/}
}

Related work

OpenTac relates to open tactile skins and to adaptive readout methods for large tactile arrays.

Tactile skins.

• FlexiTac (B. Huang and Y. Li, FlexiTac: A Low-Cost, Open-Source, Scalable Tactile Sensing Solution for Robotic Systems, arXiv:2604.28156, 2026; project page) is an open-source piezoresistive skin that estimates normal force. It comprises two flexible PCBs and a piezoresistive film sandwiched between them. Typical readout rates are around 100 Hz regardless of sparsity of contact. OpenTac measures both normal and shear forces due to our sensor design that places interdigitated row and column electrodes on a single flexible PCB beneath the film, so the load acts on the film directly, and adds bumps for shear sensing. More importantly, OpenTac-CAD provides design tools to automatically customize the sensor to any 2D or developable 3D surface. OpenTac-Xampling supports 1200 Hz+ readout rates that adapt based on the sparsity of contact, and this lets us capture contact events that occur at millisecond timescales.
• TacGooseBumps (TacGB) (W. Li, B. Yang, Y. Chen, A. Wang and M. Tomizuka, TacGooseBumps (TacGB): Retrofitting Normal-Only Tactile Sensors with Shear Encoding for Learning Contact-Rich Manipulation, arXiv:2609.34006, 2026; project page) retrofits normal-only tactile sensors with a film of elastomeric domes (e.g., cabinet bumpers). Since the sensors are retrofitted, there isn't a precise mapping from each dome to its underlying taxels. Readout rates for TacGB still depend on those of the underlying sensors they customize. For OpenTac, the normal force sensor and bump strip are co-designed in OpenTac-CAD, ensuring sub-millimeter alignment. This enables cross-sensor transfer. Further, we support vibration and pose sensing in addition to normal and shear forces, and kilohertz-rate readouts for sparse contacts.

Adaptive readout.

• ACTS (A. Slepyan, D. Li, A. Aug, S. Sankar, T. Tran and N. Thakor, Adaptive Compressive Tactile Subsampling: Enabling High Spatiotemporal Resolution in Scalable Robotic Skin, arXiv:2410.13847) samples a tactile array adaptively and reconstructs contact by sparse recovery.
• ReLACS (D. Poppert, A. Slepyan, N. Thakor and T. Tran, ReLACS: Responsive Learned Adaptive Compressive Subsampling for Efficient Readout of Large-Area Tactile Skins, Data Compression Conference 2026) learns an adaptive sampling policy for the readout of large-area tactile skins.

Both are readout methods demonstrated on other sensor hardware. OpenTac-Xampling builds on ACTS and runs in the readout-board firmware on any OpenTac skin, configured from the skin's manifest.