HiPi sensor system boosts tactile fidelity for robotic manipulation
New sensor achieves 220 Hz readout with 2048 taxels, 2x better contact geometry.
HiPi, developed by Changyi Lin and colleagues at Carnegie Mellon University, is a new tactile sensing system that combines reproducible fabrication with high-fidelity readout—two qualities that have historically been at odds in piezoresistive sensors. The system redesigns the entire hardware stack: a compact readout PCB compatible with commercial fabrication services eliminates manual soldering; a smaller STM32-based MCU module cuts cost; an optimized communication pipeline reaches 220 Hz in a bimanual setup with four dense tactile arrays (2048 taxels total); and FPCB-based conductive layers simplify sensor fabrication and stacking. Structured 3D-printed contact pattern tests show HiPi preserves contact geometry far better than a reproducible baseline: the average IoU jumps from 0.428 to 0.797, and the Dice score from 0.539 to 0.886.
The practical impact is significant. Dense piezoresistive tactile sensing has long been limited by a trade-off: easy-to-reproduce designs sacrifice signal quality, while high-fidelity architectures are hard to build and deploy. HiPi directly addresses this by making high-quality tactile arrays more accessible for bimanual manipulation and multi-fingered robotic systems. For robotics researchers and engineers, this means cheaper, faster, and more accurate tactile feedback without requiring custom fabrication expertise—potentially accelerating progress in dexterous manipulation, human-robot interaction, and tactile-based autonomous assembly.
- HiPi achieves 220 Hz readout across four dense tactile arrays (2048 total taxels) in a bimanual setup.
- New hardware stack eliminates manual soldering with a compact PCB and STM32 MCU module, lowering cost and assembly time.
- Preserves contact geometry far better than previous reproducible sensors: IoU improved from 0.428 to 0.797, Dice from 0.539 to 0.886.
Why It Matters
HiPi makes high-fidelity tactile sensing practical for bimanual robots, enabling cheaper, more scalable manipulation research.