Cross-Cutting Challenge | The Sense of Touch in Contact-Rich Dexterous Robot Manipulation
Last updated: December 1, 2025.Organizers
Lead Organizer
KAIST
Co-Organizers
Two more will be invited later
Description
Despite remarkable advances in robotic learning, perception, and control, robots still struggle with tasks that babies perform effortlessly: sliding an object to reposition it, rotating a tool in hand, or coordinating multiple fingers to manipulate a deformable item. These tasks are contact-rich and demand fine-grained sensing, interpretation, and control of forces and deformations, abilities that hinge on the sense of touch.
This Cross-Cutting Challenge (CCC) aims to bridge the persistent gap between robotic manipulation and human haptics, fostering a dialogue that unites recent trends of AI and deep learning, robot control theory, tactile sensing technology, neuroscience of manipulation, and haptic design principles. Current research in robotic dexterity often focuses on visual perception or motion planning, while tactile feedback remains underutilized due to challenges in data acquisition, representation, and integration. Yet, haptic feedback, when effectively encoded, can transform robotic interactions from open-loop trajectories into dynamically adaptive, force-aware behaviors.
This CCC will convene researchers from robotics, haptics, artificial intelligence, human motor control, and neuroscience to discuss how haptic science can accelerate progress in dexterous robotic manipulation. We will address key questions such as:
- How can tactile data (pressure maps, skin deformation, frictional states) be represented and shared across robotic platforms?
- What are effective control paradigms (e.g., impedance modulation, tactile reflex loops, learning-based policies) for contact-rich manipulation?
- How can human touch strategies inspire next-generation robotic dexterity?
- What are the data and benchmarking standards needed to make tactile learning reproducible and transferable?
Our vision is to define the roadmap for haptically intelligent robots, machines that do not merely sense contact but understand and exploit it for skillful, adaptive manipulation.
The proposed CCC builds on recent developments in both robotic manipulation (e.g., diffusion-policy-based tactile control, tactile transformers) and haptic interfaces (e.g., high-density skin sensors, soft tactile skins). By connecting these domains, we aim to inspire new collaborations, datasets, and control frameworks that bring the field closer to human-level dexterity.
Keynote Talks
TBD
Interactive Posters
TBD
Call for Individual Contributions
We seek attendee contributions to be presented as a poster or demonstration in an interactive session. To submit a poster or demo proposal, please send a title and 500 word abstract discussing how your work in haptics links to the theme of this cross-cutting challenge. We especially invite submissions from people with disciplinary backgrounds underrepresented in the haptics community. Our goal is to create an open and inclusive forum where researchers from diverse expertise can exchange insights, inspire collaboration, and spark new directions toward truly haptically intelligent robots.
Please submit your proposal by February 1, 2026 by emailing the organizers ( jhryu@kaist.ac.kr ).