LuxTac: A Flexible, Scalable, and Open-Source Optical Taxel Array with Distributed Three-Axis Force Sensing for Tactile Robot Learning
Authors: Md Rakibul Islam Prince, John Pelan, Zhiyuan Zhang, Adyaa Khera, Katherine Pesetski, Adeesh Desai, Akimizu Sugiyama, Minwoo Ki, Benjamin Alan Capelle, Temitope Ibrahim Amosa, Zijie Xu, Yihan Huang, Taehun Chung, Long Wang, Wei Li, Shaoting Lin, Yu She
Organizations: Elmore Family School of Electrical and Computer Engineering, Purdue University, West Lafayette, IN 47907, USA. · Edwardson School of Industrial Engineering, Purdue University, West Lafayette, IN 47907, USA. · Department of Robotics Engineering Technology, Purdue University, West Lafayette, IN 47907, USA. · Department of Mechanical Engineering, Purdue University, West Lafayette, IN 47907, USA. · Department of Civil Engineering, Stony Brook University, Stony Brook, NY 11794, USA. · Department of Mechanical Engineering, Stevens Institute of Technology, Hoboken, NJ 07030, USA. · Department of Mechanical Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA.
Low-cost tactile arrays have made touch a practical input for robot learning, but piezoresistive designs sense only pressure, while camera-based sensors that also resolve shear are too bulky to tile over a gripper. To solve this problem, we present LuxTac, a flexible, scalable optical taxel array, about 3.5 mm thick and built from off-the-shelf parts. Each taxel is a digital RGBC light-to-digital converter under a translucent white silicone layer capped by an opaque black film, so the back-scattered LED light falls monotonically with local indentation depth. The same board accepts two interchangeable gel cartridges: a plain one for normal sensing, and one with colored walls that encodes the local shear in the color of the returned light, with no camera, magnets, or bump overlay, making LuxTac, to our knowledge, the first optical taxel array to resolve shear at individual taxels. Characterized with an automated CNC indenter and a six-axis force/torque sensor, the normal cartridge responds linearly to force (R2 = 0.999) from 104 mN to at least 5.8 N, with 2.0% repeatability, 5.9% hysteresis, and a taxel-to-taxel variation of 6-10% after normalization. With the shear cartridge, per-taxel models estimate the three-axis force at every taxel; across three indenter shapes and nine held-out sessions, the summed estimates reach R2 = 0.96/0.96/0.92 in x/y/z, each taxel's estimate 0.92/0.93/0.94, and the median shear-direction error is 4.8 degrees. Adding LuxTac pressure readings to an Action Chunking Transformer policy substantially improves peg-in-hole success over a vision-only baseline. Design files, code, and videos are available at https://luxtac.github.io.
Figures & tables
Fig. 2: Scalability of LuxTac (renderings) . Each taxel is an independent digital detector on its own bus segment, therefore, the same taxel cell tiles into arrays of other sizes and aspect ratios without changing the readout scheme. Top to bottom: 8×8 and 8×4 arrays, and the prototype board. A US dime ( 17.9mm ) and a 1cm bar are shown for scale.
Fig. 3: Sensing principle under normal and shear loading. (a) Normal loading produces symmetric deformation, altering the optical paths between the LEDs and photodetector. (b) Shear loading produces asymmetric deformation, changing the light propagation and intensity received by the photodetector.
Sensor
Transduction
Shear/taxel
Rate
Sim.
3D-ViTac [ 2 ]
Piezoresistive
–
32Hz
–
uSkin [ 19 ]
Magnetic (Hall)
✓
100Hz
–
FlexiTac [ 3 ]
Piezoresistive
–
100Hz
✓
OpenTac [ 4 ]
Piezores. + bumps
– ∗
67Hz †
✓
SpikeATac [ 20 ]
PVDF + capacitive
–
4kHz / 40Hz
–
LuxTac (ours)
Optical, digital
✓
100–180 Hz §
✓
TABLE I: Comparison with representative tactile arrays
Fig. 4: Electronics architecture of LuxTac. (a) Exploded isometric view of the prototype’s flexible printed circuit board (FPCB) with 16 RGBC light-to-digital converters. (b) Exploded isometric view of the 4×8 FPCB with 32 RGBC light-to-digital converters. Both designs show the component assembly and four copper layers (F.Cu, In1.Cu, In2.Cu, and B.Cu). (c) Controller and data-acquisition PCB containing the Teensy 4.1, power regulation, and sensor interfacing circuitry. (d) Sensing-head component arrangement and layered tactile sensor construction.
Part
Qty
Unit
Total
Sensing pad
TCS34725FN RGBC sensor
16
6.121
97.94
White SMD LED
9
0.071
0.64
TCA9548APWR I 2 C switch
2
0.822
1.64
Passives (R, C) a
59
–
0.52
Pad total
100.74
TABLE II: Bill of materials of the prototype (unit prices in USD)
Fig. 5: LuxTac Fabrication. (a) Mode 1 (intensity-encoded), with a white translucent layer under a black opaque layer, and (b) Mode 2 (shear-encoded), with four-color silicone walls between white translucent cubes, one per taxel. Each gel is mold-cast and heat-cured on a soft acrylic layer, coated with the black opaque layer, and mounted on the FPCB layer above the photodiode array.
Fig. 6: Normal cartridge, 240 presses. (a) Normalized signal vs. force (thin: per-taxel means; black: mean over taxels; band: ±1σ across passes); s=0.02+0.170F , R2=0.999 . (b) Signal vs. depth g(δ) : per-taxel isotonic fits and the shared g ; dashed: unloading. (c) Force vs. depth, loading (solid) and unloading (dashed). (d) Force predicted from the signal, F^=h(g−1(s)) , vs. measured force, leave-one-session-out (black: binned loading mean, ±1σ ; dashed: unloading; dotted: ideal). (e)–(h) Uniformity at 0.5, 1.5, 3, and 5.5N : raw drop Δ=b−x of the pressed taxel (e) and its distribution with CV (f); normalized signal (g) and its distribution (h). Maps share one color scale per row and are drawn in the CNC frame.
Normal cartridge
Noise σ (counts; % of baseline)
1.06±0.44 ; 0.08
Detection threshold (mN)
104±49
Working range (N)
0.10–5.8 a
Saturation force (N)
>5.8 (not reached) a
Sensitivity (FS/N), linearity R2
0.170 , 0.999
Hysteresis (% FS)
5.9±1.1
TABLE III: Characterization summary of the two cartridges. Normal cartridge: mean ± std over taxels. Shear cartridge: held-out data.
Fig. 7: Shear cartridge, one model per taxel trained on all three indenters and tested on held-out sessions (each of nine sessions left out in turn). (a), (c) Per-taxel normal and lateral force vs. the label, the measured force times the share of the contact area over the taxel, for covered taxels (at least 25% of the contact; plate: cell at least half covered). (b), (d) Sum of the per-taxel estimates vs. the force/torque sensor, which does not depend on the label. (e) R2 per indenter, per taxel (solid) and summed (hatched). (f) The same model retrained with brightness only, color only, or both (per-taxel R2 , all indenters; median shear-direction error below each group).
Fig. 8: Measured and simulated readings of LuxTac for four objects in three orientations. Each grid is the mean of the 50 frames of one hold, normalized by its peak; the number is the mean cosine over all frame pairs of that grasp.
Fig. 9: Temporal tactile observations during end-to-end manipulation using LuxTac. Representative execution sequences for (Task 1) peg-in-hole insertion and (Task 2) cube sorting with smooth and dotted surfaces. Each row illustrates successive manipulation stages, including positioning, grasping, object transport, lowering, and release. The corresponding LuxTac heatmaps reveal contact-dependent responses during manipulation.
Real frames compared against
Score
Real frames of the same hold (ceiling)
0.9994
Matched simulation
0.942
Uniform pad
0.763
Simulation of a different object
0.674
Same object, different orientation
0.642
Matched simulation, taxels shuffled
0.594
TABLE IV: Sim-to-real alignment of the normal cartridge (12 grasps × 50 frames per side). Rows 1–7 are mean cosines between the 600 real frames and each comparison.
Method
Peg-in-Hole
Cube Sorting
Avg.
Vision Only
6/15 (40%) [20–64]
5/15 (33%) [15–58]
37%
Vision + LuxTac
13/15 (87%) [62–96]
6/15 (40%) [20–64]
63%
TABLE V: Real-world success rates for peg-in-hole and cube sorting tasks, with 95% Wilson confidence intervals in brackets.
Tactile sensing provides robots with contact information difficult to obtain from vision alone. Shear force is particularly important for understanding tangential contact, guiding corrective motions, and maintaining grasp stability, yet flexible and low-cost tactile arrays often measure only normal force. We introduce VecTile, a low-cost, 6 mm thin, flexible tactile array capable of three-axis force sensing. VecTile combines a piezoresistive layer with passive silicone nubs that convert normal and shear forces into distinguishable pressure patterns. The sensor can be fabricated in-house without specialized equipment. The complete system costs less than 35 USD for a 4x4 array. We validate the sensing mechanism through finite element analysis and hardware experiments. In controlled single-taxel tests over 1-5 N normal force and up to 1.5 N shear force, VecTile achieves root-mean-square errors of 0.39 N and 0.16 N for normal and shear forces. A VecTile calibrated while flat retains accuracy when curved, while a five-minute manual calibration is sufficient for practical use. Two applications demonstrate its utility. First, shear readings are sensitive enough to stop a sharp scalpel from puncturing a balloon after contact. Second, the sensor substantially improves Diffusion Policy-guided power-plug insertion. All designs and code will be released upon paper acceptance. Project website: https://jiannnh.github.io/VecTile/.
Jian Hou, Xin Zhou, Adam J. Spiers
All authors are with the Manipulation and Touch Lab, Dept. Electrical and Electronic Engineering, Imperial College London.
We present FlexiTac, a low-cost, open-source, and scalable piezoresistive tactile sensing solution designed for robotic end-effectors. FlexiTac is a practical "plug-in" module consisting of (i) thin, flexible tactile sensor pads that provide dense tactile signals and (ii) a compact multi-channel readout board that streams synchronized measurements for real-time control and large-scale data collection. FlexiTac pads adopt a sealed three-layer laminate stack (FPC-Velostat-FPC) with electrode patterns directly integrated into flexible printed circuits, substantially improving fabrication throughput and repeatability while maintaining mechanical compliance for deployment on both rigid and soft grippers. The readout electronics use widely available, low-cost components and stream tactile signals to a host computer at 100 Hz via serial communication. Across multiple configurations, including fingertip pads and larger tactile mats, FlexiTac can be mounted on diverse platforms without major mechanical redesign. We further show that FlexiTac supports modern tactile learning pipelines, including 3D visuo-tactile fusion for contact-aware decision making, cross-embodiment skill transfer, and real-to-sim-to-real fine-tuning with GPU-parallel tactile simulation. Our project page is available at https://flexitac.github.io/.
Integrating six-axis tactile feedback into robotic interfaces requires compact sensors with accurate load measurement and rapid online reconstruction. This paper presents LumiTac, a low-profile, camera-free optoelectronic tactile sensor combining spatially distributed illumination with spectral encoding. Five photoemitters and five RGBW photoreceivers are mounted on opposing circuit boards coupled through translucent silicone. Load-induced interlayer motion modulates the internal light field under compression, shear, bending, and torsion. The sensing head measures 12 mm in diameter and 4.75 mm in thickness, weighs less than 2 g, and has a component cost of approximately USD 7. A compact multilayer perceptron reconstructs three force and three moment components from 20 optical channels. Independent testing yielded coefficients of determination (R-squared) of 0.9657-0.9941, with mean absolute errors of 0.054-0.190 N for forces and 0.304-1.419 mN m for moments. Online prediction achieved an average update rate of 303.57 Hz and a median software processing latency of 0.722 ms. Experiments involving combined loading, light weights, and rapid tapping evaluated load tracking, sensitivity to small forces, and transient response, respectively. Robotic demonstrations further validated its applicability to tactile-guided manipulation and physical human-robot interaction. This study paves the way for compact, high-rate optical tactile sensing in robotic manipulation and human-robot interaction.
Hao Wu, Haotian Guo, Yu Feng +2
Department of Mechanical Engineering, National University of Singapore, Singapore