Speakers
Prof. Josh Bongard
Josh Bongard is the Veinott Professor of Computer Science at the University of Vermont and director of the Morphology, Evolution & Cognition Laboratory. His work involves automated design and manufacture of soft-, evolved-, and crowdsourced robots, as well as computer-designed organisms. A PECASE, TR35, and Cozzarelli Prize recipient, he has received funding from NSF, NASA, DARPA, ARO and the Sloan Foundation. He is the co-author of the book How The Body Shapes the Way we Think, the instructor of a reddit-based evolutionary robotics MOOC, and director of the robotics outreach program Twitch Plays Robotics.
Sim2real and real2sim
Robotics has decades of experience now with “sim2real”: training and/or designing robots in silico, and then deploying and/or manufacturing those machines in reality. What has received less attention is real2sim: what can machines “bring back” from the real world? In the case of biological robots, can sim2real’ed biobots bring back new biology? Can we task AI with designing machines that bring back as much new knowledge from the real world as possible?
Prof. Mattia Gazzola
Mattia Gazzola is an Assistant Professor in the Mechanical Science and Engineering Department at UIUC, a Blue Waters Professor at the National Center for Supercomputing Applications, and a core Theme member at the Carl R. Woese Institute for Genomic Biology. Mattia joined UIUC in Fall 2016 after a postdoc at Harvard and a PhD at ETH Zurich. Mattia's work focuses on the modeling, design, and fabrication of soft/biological architectures operating in various environments. His studies were awarded with the ETH Medal in 2013, Early and Advanced Swiss National Science Foundation Fellowships, NSF CAREER, and have been featured on the covers on several scientific journals among which Science and Nature.
Modeling, simulation and control of soft dynamic architectures
We introduce a modeling approach based on assemblies of Cosserat rods for the simulation, design and control of arbitrary muscular architectures. The obtained solver Elastica is demonstrated in a number of settings, from biolocomotion and manipulation strategies to artificial muscles and bio-hybrid systems.
Prof. John Rieffel
John Rieffel is an Associate Professor in the Computer Science Department at Union College in Schenectady, NY. His research expertise is in soft robotics, tensegrities and genetic algorithms. He received undergraduate degrees in Engineering (BS) and Computer Science (BA) from Swarthmore College, and his Ph.D. in Computer Science from Brandeis University. He has been a postdoctoral associate in Hod Lipson's lab at Cornell University, and in Barry Trimmer's lab at Tufts University.
There Are No Good Simulators for Evolutionary Soft Robotics (and There Might Never Be)
Evolutionary approaches offer a compelling tool for the design of the morphology and control of robots. Chief among them is the ability to raplidly explore design spaces relatively unfettered by preconceived human biases. Evolutionary Robotics has traditionally benefited from quality simulators, and a great deal of effort has been devoted to tackling the obstacle of "sim2real" - that is transferring evolved solutions from simulation to reality. Leveraging prior successes in rigid robots, can we ever use soft robotic simulators to evolve the morphology and control of soft robots? In this talk I'll describe some of the challenges to implementing evolutionary approaches in simulated soft robots, offer a critique of existing simulators, and then possible alternatives to existing shortcomings.
Dr. Adnan Munawar
Dr. Munawar is an Assistant Research Scientist at the Lab for Computational Sensing and Robotics (LCSR) and the Johns Hopkins University. He was previously a Postdoctoral Fellow at LCSR. He received his Ph.D. and M.S. from Worcester Polytechnic Institute (WPI) while working at the Automation in Interventional Medicine (AIM) Laboratory. He was the recipient of the Fulbright Scholarship during his M.S. program. His research interests include medical robotics, haptics and force control, simulation and animation, augmented and virtual reality, and shared teleoperation systems.
A framework for surgical robot simulation and control.
Surgical robots require precise actuation as they operate in constrained workspaces surrounded by sensitive anatomies where safety is critical. Operational safety can be enhanced by providing assistance to the surgeon and updates to the control algorithm based on situational context awareness. The Asynchronous MultiBody Framework (AMBF) is a robust platform that allows the simulation of surgical robots with complex kinematics and environments comprising soft-bodies, phantoms, and volumes generated from 3D patient scans. AMBF can also incorporate haptic devices with high-speed communication and update requirements for real-time interaction with simulated environments. AMBF is currently being used to drive physical robots with contextually aware simulation in the loop to control and correct robot trajectory and provide feedback to the surgeon(s). This talk will focus on AMBF's architecture, workflow for creating custom applications, modeling robots and environments, and recent applications.
Dr. James Bern
James Bern is a Postdoctoral Associate in the Distributed Robotics Lab at MIT CSAIL and will be starting as an Assistant Professor in the Computer Science Department at Williams College. He received his PhD in Computer Science from ETH Zurich, MS in Robotics from Carnegie Mellon University, and BS in Mechanical Engineering from Caltech. He won the ACM Siggraph Thesis Fast Forward and the IROS JTCF Novel Technology Paper Award
soft_robot.h
I will make the case that a relatively general, performant, and extensible FEM-based soft robot simulator can be written as a single ~2500 line file with minimal dependencies. The codebase compiles in one or two seconds.
Pingchuan Ma
Pingchuan is a third-year Ph.D. Student at MIT CSAIL advised by Professor Wojciech Matusik. Previously he worked with Prof. Bo Ren and Prof. Ming-Ming Cheng at Nankai University. His research mainly focuses on computer graphics and machine learning. He has been working on differentiable physics, shape optimization, and their related real-world applications.
DiffAqua: A Differentiable Computational Design Pipeline for Soft Underwater Swimmers with Shape Interpolation
We present a differentiable pipeline for co-designing a soft swimmer's geometry and controller by shape interpolation. We propose Wasserstein barycenters as a basis for the geometric design of soft underwater swimmers since it is differentiable and can naturally interpolate between bio-inspired base shapes via optimal transport. By combining this design space with differentiable simulation and control, we can efficiently optimize a soft underwater swimmer's performance with fewer simulations than baseline methods. We demonstrate the efficacy of our method on various design problems such as fast, stable, and energy-efficient swimming and demonstrate applicability to multi-objective design.
Rianna Jitosho
Rianna Jitosho received her BS degree from the Massachusetts Institute of Technology and her MS degree from Stanford University. She is a PhD candidate in the mechanical engineering department at Stanford University, is advised by Prof. Allison Okamura Stanford, and collaborates with Prof. Zachary Manchester at Carnegie Mellon University. She is a recipient of the US National Science Foundation Fellowship. Her research interests include motion planning and controls, soft robotics, and mobile systems.
Simulation and Control for Soft Growing Robots
Simulating soft robots in cluttered environments remains an open problem due to the challenge of capturing complex dynamics and interactions with the environment. Furthermore, fast simulation is desired for quickly exploring robot behaviors in the context of motion planning. In this talk, I will first introduce a model for a particular class of inflated-beam soft growing robots called “vine robots,” and present a dynamics simulator that captures general behaviors, handles robot-object interactions, and runs faster than real time. Second, I will discuss methods for fitting model parameters based on video data of a robot in motion and in contact with an environment. Finally, I will present methods for applying this simulator to control applications for vine robots.
Dr. S.M.Hadi Sadati
Dr. S.M.Hadi Sadati is a CME Research Fellow at the School of Biomedical Engineering & Imaging Sciences, King's College London, UK. He has a Bsc and MSc in mechanical Engineering from Amirkabir (2010) and Sharif (2012) U. T., and a PhD (2018) in Robotics from King’s. He has been a postdoc in Robotics System Engineering at King’s (2019-2021) and in Morphological Computation at the University of Bristol (2017-2019). He was also a Visiting Researcher at LASA, EPFL (2019 & 21), as well as Soft Robotics lab, ETH (2021), the Profs. Walker’s lab, Clemson University (2017), and Dyson School, Imperial College London (2016-2017). His research interests are soft medical robotics, morphological contribution, system dynamics, and mechatronics. Website: www.smh-sadati.com
TMTDyn: A Matlab Package for Modelling and Control of Hybrid Rigid-Continuum Robots
Hybrid rigid–continuum robot design addresses a range of challenges associated with using soft robots in application areas such as robotic surgery. Utilizing such robots poses challenges beyond standard rigid-body robots. A fast, reliable, accurate yet simple dynamic model is important to support the design, analysis, and control of hybrid rigid–continuum robots. In our recent work, we developed a modelling package for hybrid rigid–continuum systems, named TMTDyn. It utilizes four different continuum robot kinematics representations: (i) series rigid-link, (ii) piecewise constant curvature discretization (discretised Cosserat based on relative states), (iii) Finite Element Method (discretised Cosserat based on absolute states), and (iv) reduced-order shape interpolation. TMTDyn features real-time simulation, via optimized C++ models, inverse jacobian, and load compensation formulations for controller and observer designs. TMTDyn benefits from an internal domain-specific language (DSL) using Matlab’s Object-Oriented capabilities and the concept of fluent interfaces to improve validation, understandability, and maintainability of the constructed models. In this presentation, we showcase modelling a variety of continuum robots with TMTDyn such as pneumatically and tendon actuated, concentric tube, and growing robots, following by a discussion on the controller design, language implementation, the benefits, and challenges of building a Matlab-internal DSL.
Prof. Federico Renda
Dr. Federico Renda is an Assistant Professor in the Department of Mechanical Engineering at Khalifa University in Abu Dhabi, UAE. Before joining Khalifa University, he was a Post-Doctoral Fellow at the BioRobotics Institute of Scuola Superiore Sant’Anna, where he received his Ph.D. degree in 2014.
Dr. Federico Renda joined the LS2N lab at IMT Atlantique and the DEFROST Lab at INRIA as a Visiting Professor in 2018 and 2019, respectively. He currently serves as Associate Editor in the Soft Robotics and IEEE Robotics and Automation Letters journals.
His research interests include dynamic modeling and control of soft and underwater robots using principles of geometric mechanics. Dr. Renda is also a member of the Institute of Electrical and Electronics Engineers (IEEE).
SoRoSim: a MATLAB® Toolbox for Hybrid Rigid-Soft Robots
The recent growing interest in soft robotics led to significant improvements in modeling such highly deformable structures. An interesting approach is provided by the geometric variable strain methods described in [1], which is based on a strain parameterization of the soft robot. One of the most important features of the model is that it encompasses the geometric theory of rigid robotics as a particular case, making it able to generalize many of the beneficial properties of this well-established field.
In this talk, we will describe the theoretical and computational underpinning of the geometric variable-strain approach in detail. Then, a MATLAB toolbox that integrates this method will be presented through a series of validation tests and application scenarios. Together with the toolbox, the proposed model is well-posed to facilitate the modeling, statics and dynamics analysis, and low-level control of soft, rigid, or hybrid robotic systems.
[1] A. Teejo Mathew, I. Ben Hmida, C. Armanini, F. Boyer and F. Renda “SoRoSim: a MATLAB Toolbox for Soft Robotics Based on the Geometric Variable-strain Approach”, arXiv e-prints, arXiv: 2107.05494.
Prof. Minchen Li
Minchen is an Assistant Adjunct Professor at UCLA Department of Mathematics. He was a Postdoctoral Researcher in the SIG Center for Computer Graphics at the University of Pennsylvania after completing his Ph.D. in the same group, advised by Chenfanfu Jiang. Minchen is a winner of the 2021 ACM SIGGRAPH Outstanding Doctoral Dissertation Award, the 2021 Symposium on Computer Animation (SCA) Doctoral Dissertation Award, and the 2020 Adobe Research Fellowship. His Ph.D. dissertation features the invention of the Incremental Potential Contact (IPC) method, which presents a breakthrough in the notoriously challenging and long-standing problem of robust frictional contact simulation in nonlinear solid dynamics with guarantees of non-intersection, and has led to a series of follow-up works in both academia and industry.
Reliable Contact Simulation with IPC
Simulating the contact of deformable and possibly thin solids in a robust, accurate, and efficient manner is challenging. Traditionally, the contact of solids is approximately modeled with linearized geometric information near the contacting regions. This approximation is prone to generating under- or over-constrained subproblems that can produce interpenetrating or numerically unstable results, especially when large deformation of solids is also present. To avoid these issues, we propose Incremental Potential Contact (IPC) by formulating a mathematically consistent and general non-interpenetration constraint based on precisely calculated unsigned distances between boundary elements. IPC applies a customized barrier potential to directly relates the distances to the contact forces, which can grow infinitely large as the distance approaches zero to guarantee non-interpenetration. Results show reliable contact simulation using IPC even with versatile materials, large time step sizes, fast impact velocities, severe deformation, and varying boundary conditions – bringing the possibility of conveniently obtaining intricate and important dynamical details to computer graphics, computational mechanics, and robotics in a reliable way for the first time.
Dr. Sam Kriegman
Sam Kriegman is a postdoctoral fellow at the Wyss Institute at Harvard and the Allen Discovery Center at Tufts. His research draws inspiration from the origin and subsequent evolution of life, and applies the underlying mechanisms of self-organization and natural selection to the creation of novel autonomous machines. These machines can in some cases perform useful work, or they may be used as scientific tools to understand how animals evolve, grow, move, sense, and think.
Simulating xenobots and xenohybrid machines
In this talk, I will describe the computational tools used to simulate, design, and control Xenobots: biological robots composed entirely of frog (Xenopus) cells. I will conclude with ideas about how to ensure that simulated biological- and biohybrid robots (xenobots and xenohybrids) are increasingly likely to successfully transfer to reality..
Dr. Hugo Talbot
...
Moritz Alexander Graule
I'm a PhD student at Harvard's Microrobotics Laboratory. I'm interested in the computational design, simulation, and control of soft robots that interact with their environment in a robust and safe manner; and applying deep learning to improve clinical workflows and patient outcomes.
SoMo and SoMoGym: fast simulations with sufficient accuracy
While the inherent compliance of soft robotic systems lets them passively adapt to variable environments and operate safely around humans and fragile objects, it also provides a fundamental challenge in simulating these systems: their high number of degrees of freedom means that accurate simulation is computationally expensive, and therefore slow. Our frameworks SoMo and SoMoGym leverage rigid-link approximations of soft robots for fast simulations that achieve sufficient accuracy at the system level for design exploration and controller development.
Dr. You Wu
Dr. You Wu is a robotics evangelist. He designed and deployed soft robot commercially before, and now he is promoting best practices in robot development processes on behalf of MathWorks. Dr. Wu received his Ph.D. degree from MIT, and his thesis is about the design of soft-material robot for water pipeline inspection. He commercialized his Ph.D. thesis and launched Watchtower Robotics, a startup that deployed those robots into municipal water pipe networks. Currently, Dr. Wu is the Robotics Industry Manager at MathWorks, promoting tools and workflows to accelerate the development of real-world robotic applications.
What We Learnt From Deploying Soft Robot in Commercial Applications
With the rise of robotic startups led by young Ph.D.s in recent years, we hear a lot of personal experiences in the difference between developing a robot that works in a lab and developing one that works in real world. What does the difference look like in the soft robot space? Dr. You Wu is here to share his story designing and deploying soft robots that went inside and inspected city underground water pipes. He is also going to discuss his “I wish I knew” list, which summarizes the good and bad practices he learnt from his own experience and other robotics companies’.
Dr. Robbie Balcombe
Robbie works as the Technical Director at COMSOL UK. He graduated from the University of Strathclyde in 2007 with a MEng in Aero-Mechanical Engineering, and then carried out his PhD in the area of numerical modeling of rolling contact fatigue at Imperial College London.
Simulating Soft Robotics with COMSOL Multiphysics®
As with many emerging technologies soft robotics is an area where engineers and researchers are leveraging computational models to help answer key design questions. Simulating soft robotics presents some specific challenges because of a need to couple together different physical effects, such as electric fields or magnetic fields with structural mechanics, for implementing control methods and sensors, and also the compliant nature of the materials involved which means that their behavior is very often complex and nonlinear.
COMSOL Multiphysics® is well placed to address these specific challenges and during this presentation we will discuss different nonlinear structural material models and ways that physics can be coupled together for soft robotics simulations.
Prof. Stanislao Grazioso
Stanislao Grazioso is Associate Professor in the Department of Industrial Engineering at University of Naples Federico II, Naples, Italy. He holds a Bsc (2012) and a Msc (2014) in Mechanical Engineering and a PhD (2019) in Industrial Engineering from University of Naples Federico II. He has been a visiting researcher at the George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology (Atlanta GA, USA) and at the Department of Aerospace Engineering, University of Maryland (College Park MD, USA) in 2016. He received the Georges Giralt PhD Award in 2019. His current research interest include: modeling and simulation of articulated and continuum soft robotic systems, design and fabrication of soft robots.
SimSOFT: a physics engine for hybrid soft robots
SimSOFT allows the modeling and simulation of hybrid soft robots, namely robots composed by rigid and/or soft continuum links interconnected by rigid and/or elastic joints in serial/parallel kinematic topologies. In this talk we will describe the modeling approach behind SimSOFT, which is based on the Cosserat rod theory and the finite element method in a geometric mechanics framework. Then, we will show its graphical user interface and we will present some use cases developed in the recent years.
Dr. Andrew Spielberg
Andrew Spielberg is a Postdoctoral Fellow at Harvard University. His research focuses on developing algorithms to co-design novel types of rigid and soft robots in form and behavior, and automatically fabricate them. His work has touched upon topics in soft matter and differentiable simulation, numerical optimization and machine learning for robot control and design, and digital fabrication processes such as 3D printing and textile-manufacturing. He received his B.S. and M.Eng from Cornell University and his PhD from MIT. His work won a best paper award at CHI, and has been nominated for best paper awards at ICRA and Robosoft.
Differentiable Simulation: A Workhorse for Computational Soft Robotics
Computational soft robotics promises algorithms that will aid us in controlling and designing soft robots with the complexity of soft organisms found in Nature. We need, however, the right tools to power these algorithms --- namely, simulation methods that are fast, differentiable, and physically-based. In this talk, I will discuss recent work on differentiable soft robot simulators such as ChainQueen and DiffPD, whose fast differentiability unlocks applications in optimal motion-planning, computational co-optimization over materials, sensing, actuation, shape, and control, and tools for overcoming the sim-to-real gap. Comparison to previous work will show how these "right tools" vastly outperform the "wrong" ones and unlock novel capability. Beyond current applications, part of this talk will look to the future; with some of the low-hanging fruit in differentiable soft robot simulation now plucked, what new capabilities are needed in order to solve the most salient problems with real-world impact?