DigiTac: A DIGIT-TacTip Hybrid Tactile Sensor for
Comparing Low-Cost High-Resolution Robot Touch

Nathan F. Lepora,  Yijiong Lin, Ben Money-Coomes, John Lloyd
Department of Engineering Mathematics & Bristol Robotics Laboratory,
University of Bristol, UK

Abstract: Deep learning combined with high-resolution tactile sensing could lead to highly capable dexterous robots. Here we customize the DIGIT low-cost high-resolution tactile sensor (Facebook AI Research and GelSight) to have a 3D-printed sensing surface based on the TacTip family of soft biomimetic optical tactile sensors. The DIGIT-TacTip (DigiTac) enables direct comparison between these distinct tactile sensor types. We introduce a tactile robot system comprising a desktop arm, mounts and 3D-printed test objects. We use tactile servo control to compare the DIGIT, DigiTac and TacTip for edge- and surface-following over 3D-shapes. All three sensors performed similarly at pose prediction, but their constructions led to differing performances at servo control, offering guidance for researchers selecting or innovating tactile sensors.

Left: the DigiTac mounted on an affordable desktop robot arm (Dobot MG400). Right: close-up of sensor and a captured high-resolution tactile image.

IEEE RA-L paper

arXiv preprint

DigiTac Design

Test
Robot

Servoing
Code

Data

DigiTac design

TacTip sensing surface integrated with DIGIT base.
A) 3D-printed TacTip, B) acrylic window, C) lighting PCB, D) plastic housing, E) camera PCB, F) back housing.

DIGIT

DigiTac

TacTip

Compare via Tactile Servoing – Edge Following

Compare via Tactile Servoing – Surface Following

Conclusions

• Adapted the DIGIT to have a TacTip sensing surface (DigiTac)

• DIGIT, DigiTac, TacTip performed well at edge and surface pose prediction; typically, ~0.1mm and ~1deg accuracy.

• Compared on tactile servo control tasks:
  1. Edge following: Similar performance
  2. Surface following: Only DigiTac and TacTip could complete

• The main difference between the sensors is that the DIGIT sensing surface is flat and fairly inelastic compared to the soft curved sensing surface of the TacTip, which was more suited to servo control tasks.

Open-source Robotics

github.com/nlepora/digitac-design

bitbucket.org/nlepora/tactip-hardware

github.com/nlepora/pbts-2d

Datasets – data-bris: tactile-servoing-2d-dobot

See also

Tactile Gym 1.0

Tactile Gym 2.0

Citation

@article{lepora2022digitac,
  title={DigiTac: A DIGIT-TacTip Hybrid Tactile Sensor for Comparing Low-Cost High-Resolution Robot Touch},
  author={Lepora, Nathan F. and Lin, Yijiong and Money-Coomes, Ben and Lloyd, John},
  journal={IEEE Robotics and Automation Letters}, 
  year={2022},
  doi={10.1109/LRA.2022.3190641}
}