Physical AI
Multi-modal Robot Learning
Embodied Tactile Intelligence
About
us
Who we are
RoboTac is an interdisciplinary research laboratory dedicated to advancing Physical AI through Multimodal Robot Learning and Embodied Tactile Intelligence.
Our research focuses on developing intelligent robotic systems that can perceive, understand, predict, learn, and act through physical interaction with the world. Inspired by human sensorimotor intelligence, we investigate how robots can integrate multiple sensory modalities, particularly vision, touch, and proprioception, to build meaningful representations of their environments, learn from experience, and adapt their behavior to dynamic and unfamiliar situations.
Our goal
At the core of our research is the understanding that physical intelligence emerges not only from observing the environment but also from actively interacting with it. We therefore investigate active, interactive, and predictive perception, enabling robots to explore their surroundings, interpret physical contact, anticipate interaction outcomes, and continuously refine their internal representations through experience.
Our work
Our work combines multimodal perception and robot learning with world models, foundation models, and embodied intelligence to advance adaptive robotic behavior. Particular emphasis is placed on tactile-driven perception and learning, contact-rich and dexterous manipulation, whole-body tactile intelligence, and safe physical human–robot interaction.
Beyond conventional robotic platforms, we explore how these principles can enable intelligent systems across different embodiments and application domains, including humanoid robotics, industrial automation, and human-centered intelligent systems. We also investigate energy-efficient computing approaches to support the deployment of intelligent perception and learning in real-world environments.
By bridging fundamental scientific research with practical applications, RoboTac aims to advance a new generation of physically intelligent systems that learn through interaction, reason about the physical world, and adapt their actions to changing environments.
Latest News
Prof. Kaboli to Deliver Keynote at the Artificial Intelligence Conference
Prof. Kaboli to Deliver Keynote at the Artificial Intelligence for Medicine Conference
Prof. Kaboli will deliver a keynote talk at the Artificial Intelligence for Medicine Conference, organized by Charité – Universitätsmedizin Berlin and the Berlin Institute for the Foundations of Learning and Data (BIFOLD), on November 24th.
Her presentation will highlight recent advances from our lab and provide new perspectives on cutting-edge research in AI-enabled multimodal perception and learning for robust in-vehicle vital-sign monitoring.
We look forward to her contributions to this important event and to engaging with the broader research community.

Associate Editor, IEEE ICRA 2026
Excited to share that Prof. Mohsen Kaboli will be serving as an Associate Editor (Manipulation & Grasping and Mechanisms, Design, and Control ) for IEEE ICRA 2026! Looking forward to contributing to the review process and engaging with cutting-edge robotics research.

New paper accepted at IEEE Robotics & Automation Letters
We’re excited to share that our paper, “Tacser and Action-Conditioned Latent Filter for Generalizable Robotic Surface Perception” by Anirvan Dutta, Yerkebulan Massalim, Etienne Burdet, Mohsen Kaboli, has been accepted in IEEE Robotics and Automation Letters (RAL), Oct 2025.

Disclaimer: The content of this website may be downloaded / printed for personal / public use, and reproduction of the texts is permitted with appropriate acknowledgement. The website dose not reflect the views of the employers and only advertises already publicised contents.

