Live Program

Dynamics of continuum and soft robots: a strain parametrization based approach

Frederic Boyer

Institut Mines Telecom Atlantique, France

Abstract: In this talk, we propose a new dynamic model of Cosserat beams in view of its application to continuum and soft robotics manipulation and locomotion. In contrast to usual approaches, it is based on the non-linear parameterization of the beam shape by its strain fields and their reduction on a functional basis of strain modes. While remaining geometrically exact, the approach provides us with a minimal set of ordinary differential equations in the usual Lagrange matrix form that can be exploited for analysis and control design. Inspired from rigid robotics, the calculation of the matrices of this Lagrangian model is performed with a new reduced inverse Newton-Euler algorithm. To assess the approach, this Lagrangian model is compared against a well-validated finite elements method through several benches of non-linear structural statics and dynamics.

Poster Session I 14:30 – 15:00

•Conformation Control of a Soft Manipulator

•Hybrid Rigid-Soft Robots: Design, Modeling, Perception, and Control

•Predicting Behaviours of Fluid Elastomer Actuators

Model-based control of soft robotic systems

James Bern

ETH Zurich, Switzerland

Abstract: Unlike their rigid cousins, soft robots conform naturally to the environment. This makes them inherently safe and robust to uncertainty, and therefore promising for applications requiring close interaction with people, such as elderly care, medicine, and rehabilitation. However, these benefits are not for free. The same softness that makes soft robots so exciting also renders them unintuitive to design, and difficult to control. For example, to answer the seemingly simple question of calculating a soft robot's inverse kinematics, we must account for highly nonlinear soft body physics and contact dynamics. A fundamental challenge in soft robotics therefore is to make it easy to predict and control soft robot motion. We propose a differentiable soft robot simulator based on the Finite Element Method. I will show how to leverage this simulator to answer not only the classic questions of forward kinematics and inverse kinematics for soft robots, but also to start answering the harder questions of soft locomotion, soft manipulation, and more.

Fast FEM for soft-robots: a bridge between robotics and structural mechanics

Christian Duriez

INRIA, France

Abstract: During this presentation, I will briefly present the results obtained by the DEFROST team on modeling and control of soft-robots and show that these results create a bridge between robotics and structural mechanics. With the increase of computational power and using adequate methods, we can use these models for robotic applications. An important message is that using well known methods for mechanical modeling for deformable solids, soft robotics community will make a step further in maturity and will get closer to the present challenges faced in today in robotics.

Design and Control of Soft Robotic Characters

Moritz Bächer

Disney Research, Switzerland

Abstract: As the field of Soft Robotics matures, the complexity of the tackled problems will inevitably increase, and computational tools for simulation and optimization will become key aspects of soft robot design and control. In this talk, I will highlight how computation paves the way toward industrial-grade robots that are lightweight and inexpensive, yet functional. Specifically, I will discuss how we can leverage differentiable simulation to (1) accurately characterize soft robotic materials, (3) computationally control very soft robotic systems while suppressing visible mechanical oscillations, and (3) automate the design of proprioceptive soft robots with desired functionality.

Poster Session II 17:00 – 17:40

•Kinematics and Shape Sensing of a Collaborative Continuum Robot

•Control and Trajectory Optimization for Soft Aerial Manipulation

•Introducing FlowIO – Modular, Integrated Development Platform for Soft Robotics

•A discrete geometric approach to simulation of soft robots

Modeling Challenges and Applications of Soft Robotics: Sensing and Control

Nabil Simaan

Vanderbilt University, USA

Abstract: The talk will discuss modeling and control of soft robots with emphasis on applications as steerable implantable devices. Simplified approaches for modeling and control of steerable slender continuum devices made of elastomers and their possible applications for cochlear implant surgery will be used to motivate this talk. The talk will also discuss how some of these approaches can inform indirect sensing and steering of electrode arrays for cochlear implant surgery and other applications. Recent extensions of these approaches to contact detection using pneumatic braided muscle actuators will also be discussed along with challenges and suggestions for future research directions.

Sampling-based Optimal Control with Learned Models for Soft Robots

Marc Killpack

Brigham Young University, USA

Abstract: Model-based optimal control of large-scale soft robots is difficult for many reasons. Two of the main limitations are the difficulty of modeling such systems, and the tractability of the optimal control problem given the large number of degrees of freedom. First principle-based models can be difficult to derive and validate for soft robots. In addition, if the form of the soft robot platform changes, the method to model the platform may also need to change drastically. While any attempt to include additional states to improve the model accuracy also increases the complexity and decreases the tractability of the optimal control problem. In this talk, with an eye towards enabling multiple soft robots to coordinate motion for co-manipulation tasks, we present two approaches to solve these problems. First, we present learned models and DNN architectures that enable closed-loop optimal control of large-scale soft robots. Second, we present a sampling-based method to find high-quality solutions in a model predictive control formulation. Finally, we also present the limitations of this work and needed changes to reach the full potential of model-based control for soft robots.

Poster Session III 14:00 – 14:30

•Observer Design for a Fabric-Reinforced Inflatable Soft Robot

•Soft Robot Workspace Expansion Via Smart Material

•Gaussian Process Models for Soft Robot Locomotion

Continuum Robot Modules and their Control

Ian Walker

Clemson University, USA

Abstract: The talk discusses issues in the control of continuum robots. Beginning with an introduction to core continuum robot modules, or sections, examples of alternative types of controllers proposed for them are presented and discussed. Questions regarding which type and complexity of controllers are most appropriate for compliant continuum robotic structures are introduced and addressed.

Hybrid Dynamics for Variable-length Multisection Continuum Arms

Isuru Godage

DePaul University, USA

Abstract: Variable-length multisection continuum arms have unique features such as inherent compliance that make them suitable as collaborative robots. These unique features also render the control of continuum arms challenging. Despite being around for close to two decades, they are yet to leave the laboratory settings. A key reason for this lag is the lack of real-time dynamic control. To this end, efficient dynamic models are indispensable. The dynamic models that have been proposed so far are limited either in modeling accuracy or numerical efficiency to be applied in dynamic control. In this talk, we present a novel hybrid dynamic modeling approach for variable-length multisection continuum arms. Thus, unlike prior lumped or discrete models, the proposed model does not betray the robot's continuous and smooth bending and thus produce accurate dynamics results. The model is then implemented using the composite rigid body algorithm (CRBA) and incorporates continuum arm kinematics constraints. The algorithm delivers 26 times faster simulations than real-time. The model accuracy is verified against the state-of-the-art dynamic models and experimentally validated. This is the first sub-real-time dynamic model that uses CRBA based on a continuous kinematic model for variable-length multisection continuum arms.

Break 15:30 – 15:40

Personalized Stents and Robotic Balloon Endoscopes for Airway Management

Pierre E. Dupont

Harvard Medical School, USA

Abstract: In medical interventions, there are advantages to using soft materials for both implanted devices and therapeutic robots. This talk explores their use in treating blockages of the airways for which two fundamental procedures are stenting and balloon dilation. An in vivo molding technique will be presented which enables patientsto receive a stent tailored to the size and shape of their airway. In this approach, a soft stent with a liquid UV-curable core is balloon expanded so that it conforms to the airway walls and is then quickly cured to this shape. A soft robot will also be presented that can navigate through the airways serving as both an endoscope and balloon dilator. Combining these functions which are normally performed using two instruments avoids the visual occlusion problems encountered during balloon dilations. Fabricated using inexpensive materials and processes and powered by pressurized air and vacuum sources that are found in standard medical operating rooms, the design lends itself to use as a disposable

Modeling and Control for New Continuum Structures

Caleb Rucker

University of Tennessee, USA

Abstract: This talk will review modeling for some new compliant robotic structures. First, a new actuation concept - concentric precurved bellows - will be described and demonstrated, including simple constant curvature models, and some recent constant-curvature modeling discussions in the literature will be highlighted. Next, general kinetostatic problem formulations will be categorized and discussed. Recent work shows that model-based estimation of continuous load distributions along soft robots is feasible and useful. Finally, the talk will highlight real-time dynamic simulation work for all types of continuum and soft robots, including concentric-tube robot snap-through behavior.

Localization and Control of Growing Robots in Medical Applications

Tania Morimoto

University of California San Diego, USA

Abstract: Soft, growing robots achieve locomotion by material extending from their tip. They are inherently compliant and can safely navigate through tight turns and highly constrained environments that prove challenging for traditional robots. Despite the potential of these robots, there remain a number of challenges to practical implementation, particularly related to shape estimation and control, that must be addressed. In this talk, we present a low-cost, wireless, permanent magnet-based method for localizing the tip of these robots. A permanent magnet is placed at the robot tip, and an array of magneto-inductive sensors is used to measure the change in magnetic field as the robot moves through its workspace. We develop an approach to localization that combines analytical and machine learning techniques and show that it outperforms existing methods. In addition, we explore the potential to use this localization method for closed-loop control of a steerable growing robot.

Panel discussion 17:10 - 18:00