Toward Autonomous Continuum Robots: Modeling, Estimation, and Interaction

Prof. Jessica Burgner-Kahrs

Professor, University of Toronto with appointments in Computer Science, Mechanical & Industrial Engineering, and Mathematical & Computational Sciences

Location: Mechanical Engineering Research Lab, 203 (link)

(Hosted by Allison Okamura)

Abstract

Continuum robots enable access to confined, contact-rich environments that remain inaccessible to conventional rigid-link systems, including pipes, turbomachinery, and other critical infrastructure. Their compliant, continuous structure provides inherent advantages for safe interaction and adaptation, but also fundamentally changes how robots must be designed, modeled, and controlled. This talk presents a perspective on continuum robotics that integrates physical embodiment with uncertainty-aware computation. It highlights the role of physical intelligence in shaping interaction, locomotion, and manipulation capabilities, focusing on tendon-driven continuum robots. Morphology, material distribution, and modularity are shown to encode task-relevant behavior for operation in constrained environments. Modeling approaches are then discussed, including physics-based simulation frameworks that capture the coupling between actuation, elasticity, and environmental interaction. While essential for prediction and control, these models are limited in practice by unmodeled effects, parameter uncertainty, and sparse sensing. To address this, a stochastic framework for state estimation is introduced that formulates shape reconstruction as a continuous inference problem. By leveraging an analogy to continuous-time trajectory estimation in mobile robotics, deformation is modeled as a Gaussian process over arc length, enabling principled fusion of sparse, noisy measurements with prior assumptions and yielding full-state estimates with uncertainty. Finally, the talk outlines how uncertainty-aware state estimation enables higher-level capabilities such as contact-aware planning and exploration in constrained environments. Together, these results support a unified view in which embodiment and probabilistic computation are jointly leveraged to enable autonomous continuum robotic systems.

Bio: 

Jessica Burgner-Kahrs is Professor at the University of Toronto with appointments in Computer Science, Mechanical & Industrial Engineering, and Mathematical & Computational Sciences. She is the founding Director of the Continuum Robotics Laboratory. She received her Ph.D. in computer science from Karlsruhe Institute of Technology in 2010 and held prior positions at Vanderbilt University and Leibniz University Hannover. Her work has been recognized with the Heinz Maier-Leibnitz Prize and the Lower Saxony Science Award, and she was named a Young Global Leader by the World Economic Forum in 2019.