A New Approach in Robotics: ACROSS Connects Tactile Sensors

L3S Best Publication of the Quarter (Q1/2025)
Category: Robotics

ACROSS: A Deformation-Based Cross-Modal Representation for Robotic Tactile Perception

Authors: Wadhah Zai El Amri, Malte Kuhlmann, Nicolás Navarro-Guerrero

Presented at IEEE International Conference on Robotics and Automation (ICRA)

The paper in a nutshell:

Our research presents ACROSS, a method that enables the transfer of information between different robotic tactile sensors by modelling how they deform when interacting with objects. This framework allows older datasets from discontinued sensors to be used with modern ones, saving time and resources or exchanging results between labs using different sensors more easily.

Which problem do you solve with your research?

This research brings us closer to an invariant tactile representation, which is yet lacking for tactile sensors. As one possible application, we can reuse older datasets into alternative sensors, thus saving time and resources.

What is new about your research?

Unlike existing approaches, ACROSS converts data between entirely different sensor types by modelling how they physically deform when touching objects. This approach allows low-resolution sensor data to be translated into high-resolution images, something not done before.

What is the potential impact of your findings?

ACROSS makes robotics research more efficient by reusing old datasets, reducing costs and effort in data collection. It also enables different research groups to share and compare data, speeding up advancements in robotic touch technology.

Paper link: https://arxiv.org/abs/2411.08533