HRI Lab Pusan National University

Research

Body-scale robotic skin on a robot arm, with an exploded view of the multilayered fabric and the multimodal sensing module (capsule, vented screw, microphone)

Fabric-based robotic skin using multimodal sensing modules

Jul. 2022 – Jan. 2024

  • A robotic skin designed with multimodal sensing modules (MSMs) and multilayered fabric to have two complementary transducing mechanisms.
  • The complementary sensibilities improve classifying 10 touch types.
  • The MSM integrates two types of transducer and can be easily made on rigid surfaces with its vented screw, simplifying a fabrication.
  • The multilayered fabric is light and flexible to easily conform on three-dimensional surfaces.
  • The skin fabricated on a robotic arm to demonstrate physical human-robot interaction by reacting differently to various touches.

Related publications ICRA 2023 T-RO 2025

Biomimetic robotic skin compared with human skin layers, its pressure and vibration sensing principle, large curved skin, and repair after damage

Repairable biomimetic elastomeric robotic skin

Mar. 2021 – Jun. 2022

  • A skin-inspired multi-layer structure is fabricated with silicone elastomer and ionic hydrogel to transmit alternating electric currents and vibrations in response to touch.
  • Tomographic imaging methods are utilised to reconstruct a multi-modal tactile image from measurement data.
  • The skin can be repaired even after severe damage, such as an incision, using chitosan topohesive and silicone adhesive.
  • A sensorised prosthesis is made as a proof-of-concept.

Related publications Sci. Robot. 2022

Robotic skin with embedded microphones on a UR5e robot arm reacting to a pat, and a large cylindrical skin localising and classifying a scratch

Robotic skin using microphones for dynamic touch perception and touch classification

Jan. 2019 – Nov. 2022

  • Microphones are sparsely distributed beneath a textured fabric to collect vibrations elicited by a dynamic touch.
  • Collected vibrations are processed with an analytical decay model to localise a touch and achieve a super-resolution.
  • Spatiotemporal properties of signals are analysed with CNN to classify touch types (pat, rub, etc.).
  • The skin is made on a large cylindrical surface and robotic arm to demonstrate scalability and physical human-robot interaction application.

Related publications ICRA 2021 T-MECH 2024

32-channel EMG electrodes on the hand and forearm, sEMG sensors on a lower-limb exoskeleton, signal-processing steps, and a classification plot

Intention and environment recognition using surface electromyography (sEMG)

Mar. 2018 – Dec. 2021

  • sEMG is used for various recognition tasks in human-machine interfaces by its early detection compared to physical sensors.
  • Different forms of sEMG and signal processing methods are implemented depending on the subject and target application.
  • For assistance in a healthy subject, single-channel electrodes are attached to multiple muscles to recognise gait environment and motion intention.
  • For hand motion rehabilitation in stroke patients, high-density electrodes are used to detect an intention in binary classes from poor signal quality.

Related publications FITEE 2019 ICORR 2022 EMBC 2023