Workshops Day (4 April)

The field of soft robotics has experienced tremendous growth over the past decade, drawing interest due to the utility of soft robots in applications involving handling of fragile objects or close collaboration with humans, their relatively low cost and light weight, and their resistance to impact and harsh conditions. To fully realize these advantages, researchers are now investigating methods to simplify or replace the existing hard, bulky control infrastructure ubiquitous to soft robots with a combination of onboard control methods and advances in external software control. Research in this area has been fast-paced, with many concurrent and complementary developments in just the past three years, but opportunities for further advances remain. This workshop will bring together a diverse group of researchers investigating the simplified control of soft robots to identify promising future directions.

Recent advances in materials and manufacturing approaches have opened new possibilities for sensor design in the context of soft robotics. Inspired by the remarkable performance of biological sensing systems we believe that understanding biological solutions can provide useful design guidelines to develop bio-inspired soft sensors. However soft sensing is still far from allowing natural-like perception. This is mainly due to the multiple aspects that need to be considered in the design including materials and structure the implementation of an appropriate computational layer (neurons) and their interplay with sensed environment. In this workshop we aim to discuss how the lessons from biology in morphological sensing may inform and benefit feasible implementation of embodied sensing in soft robotics. Besides the presentation the workshop will feature in-depth panel discussions among biologists, roboticists, neuroscientists, and material scientists about current challenges and ways towards a more holistic pipeline from biological system to abstraction to practical soft robotics applications.

There has been a long-standing philosophical debate about the relationship between body and mind. Today this debate still powers a profound scientific desire to deepen our understanding of the nature of both animals’ and machines’ intelligent, adaptive behavior. To gain further insights into intelligence and explore how our brain and whole selves develop through physical interactions with the world, Embodied Intelligence places the physical entity of the human body at the center of this subject. In the age of AI and Machine Learning, Embodied Intelligence research remains highly important as it can deliver valuable input which enhances the impact of conventional AI technology.

Soft Robotics research has recently shown how conceptual issues of Embodied Intelligence can be turned into physical reality. Only by promoting the soft robotics technologies towards a more Embodied Intelligence framework, will the technologies reach the next level of autonomous adaptive systems.

The proposers for this workshop also annually organize the International Conference on Embodied Intelligence (https://embodied-intelligence.org/) conference to discuss these various challenges. Last year, the conference attracted over 100 speakers and 1000 participants, reflecting both the importance and the necessity of this research discussion within the collaborative community.

This workshop aims to continue the stimulating discussions in the workshop, and strengthen the participation of soft robotics researchers.

Soft Robotics is a trending multi-disciplinary research topic inspired by highly dexterous and deformable biological bodies in the form of intrinsically soft robotic platforms. However, the compliance offered by soft robots has disadvantages resulting in modeling, design optimization, system analysis, control, and automation challenges hindering their real-world deployment. Developing software frameworks for addressing the aforementioned needs in Soft Robotics research is receiving increased attention recently. Such frameworks should be easy to use to be widely accepted by the multidisciplinary Soft Robotics research community that gathers researchers from different expertise and backgrounds. This workshop brings together the academic and industrial viewpoints on the requirements and ways to address the grand challenge of achieving a unified software framework for Soft Robotics Research.

The main aim of the workshop is to introduce the already available toolkits to the wider community and to inspire new approaches in developing such platforms for soft robots with the application and user experience in mind. We will bring together recognized experts in both modeling and control of soft robots, trying to answer questions such as:

What are the impacts of such frameworks?

What are the immediate and most general requirements in the Soft Robotics community to be addressed by a soft robotic platform?

How to decide between existing soft robotic toolkits for a given simulation, analysis, or control task?

How to design an easy-to-use toolkit to be widely accepted for multidisciplinary Soft Robotics research?

What are the means of achieving fast yet accurate computational performance for a variety of tasks such as modeling, design optimization, controller design, and deep-learning research?

How to achieve inter-operable platforms compatible with standard platforms in the community such as (e.g. C/C++, Python, Matlab, and ROS?

What problems can not be solved with a ‘perfect’ simulator?

The field of Soft Robotics is broad in terms of actuation technologies from pneumatic/fluidic, or thermal, or electrical, or chemical to mechanical domains, and from the microscopic to the macroscopic scale. Applying these actuation technologies in real world application comes with the challenge of developing energy-efficient systems. Energy is the lingua franca of engineering, and an energy-based approach has the potential to define a single approach that highlights the key features of a system. This workshop aims to educate researchers working on soft robotic systems on energy-based approaches specifically applied to the systems they are developing by using concepts such as the port-Hamiltonian approach and bond-graph theory. Several examples of these approaches will be provided through case studies presented by the speakers of this workshop.

To model, or not to model? That is a question in soft robotics. Is modeling the way towards next- generation soft robot design, control and best use of embodied intelligence? To what extent can we model the interaction of a soft robot body with the environment and fully grasp the essence of embodied intelligence? Can we achieve digital twins of our soft robots? What is the required relation between accurate modelling and accurate fabrication? How to overcome the reality gap? Those and other questions will be discussed at the workshop. We argue that a wider adoption of computational modeling in soft robotics provides opportunities for progressing towards a model- informed discipline, ultimately responding to societal needs. The challenges involved stand as research opportunities in multiple fields. Transition from prototype-based to model-informed design in soft robotics is within grasp, in an interdisciplinary dialogue that we wish to promote starting with this workshop.

In recent years, the field of soft aerial robotics has benefitted from the utilization of reconfigurable, flexible, soft, and morphologically adaptive structures. This workshop highlights the current and future utilization of soft and morphologically adaptive structures in aerial robot related applications: such as  (1) improving maneuverability and flight efficiency; (2) multi-modal mobility across terrain interfaces and fluid boundaries; (3) robustness to landing and collision (4) manipulation, perching, resting, and energy management in complicated environments; (5) bio-inspired aerial construction and nesting; (6) bio-hybrid flying structures, biodegradability; (8) self-regeneration and self-healing. The workshop aims to host the leading researchers in the fields of bio-inspired, reconfigurable, morphing, and soft aerial robots to create a dialogue on state of the art, identify technical/conceptual barriers, and outline key challenges and opportunities for soft aerial robotics. The workshop will outline instances where inspiration from biology is utilized to develop the next generation of aerial vehicles. Highlighting various enabling technologies and system features desired in novel aerial robots, that can perform multi-capabilities, multi-modal tasks, and autonomously operate in real-world conditions.

State-of-the-art soft robots that can burrow into soil or granular media are limited. Due to sub-terranean forces acting on the robot body, enabling burrowing soft robotic applications is highly challenging and requires deeper research. Nature provides many examples of successful burrowing mechanisms, exhibited by organisms as varied as worms, bivalve mollusks, and plants. How can we learn from biology and create more capable soft burrowing robots? By analyzing soft biological burrowing mechanisms, and identifying the grand challenges we need to overcome, we can manufacture next-generation soft burrowing robots. In this workshop, we will convene experts across disciplines: physics, biology, robotics, mechanical engineering, and materials science to discuss the grand challenges and how to solve them. Furthermore, with an additional speaker from industry, this workshop will also highlight how the advantages of soft robots can be harnessed for burrowing applications, minimize disruption to the environment while digging and real-world use cases.

The design of soft robots involves many choices spanning materials, actuators, sensors, and geometric and force considerations. Often, these choices are made heuristically with limited or no systematic framework. A heuristic-based design approach can lead to significant time spent iterating on hardware, incur extra costs, material waste, leading to underperforming robots. This workshop will highlight existing design optimization (inverse design) methodologies for soft robots and provide a forum for discourse on approaching soft robotics with systematic design thinking. Key questions include: Are the same design pipelines for conventional mechanisms applicable to soft robots? Can we leverage knowledge from other domains for soft robotics? What are trade-offs in deterministic vs. non-deterministic methods? How can we mitigate the Sim2Real gap?

In his 2018 book “Where’s my flying car?”, the author J Storrs Hall asks why, some 60 years later, the future envisioned in the 1960s has never really come to materialize. Where for example, are the fully interactive robots that were supposed to walk amongst us? Why instead, do we hear about Teslas running into the back of parked fire-trucks on the highway, or Waymo cars getting stuck in front of a cone in a construction zone until a service operator arrives and gets it unstuck? The reason is simple: Both the fields of “artificial intelligence” and robotics have been taken hostage by a litany of unquestioned intellectual hegemonies that rule supreme. The solution therefore requires a complete and total reframing of the problem at hand. More concretely however, the “Alternative AI” framework calls for the creation of synthetic organisms, replete with fully respecting the lineage of how natural intelligence arises, and a strong and rigorous rejection of a multitude of harmful ideologies that have thus far held us back. 

Following on the successes of our 2018 and 2019 workshops, the purpose of this workshop is to provide a forum to discuss recent advances and challenges in the field of tensegrity robotics. As compliant structures with tunable stiffness, tensegrities are a compelling platform with which to study the spectrum of robotic morphologies, from soft to rigid. Tensegrities are relatively simple to design and fabricate, and yet they present all the same challenges and pathologies of more conventional robots. Most valuably, they are incredibly modular, requiring few distinct parts, and allowing them to scale quite well in complexity. This workshop will provide participants with an opportunity to present and discuss the current state-of-the-art in tensegrity robotic research, through a series of invited talks, a solicited poster session, and an interactive panel discussion.

This workshop will provide a series of in-depth tutorials on how to use the open source simulation framework SOFA and the SoftRobots plugins made by the DEFROST research team to model, simulate and control deformable robots. At the end of the tutorial session, we expect attendees to be capable of modeling and controlling basic soft robots with SOFA including being able to precisely model contact with inverse control. A special moment will be allocated on the new SofaPython3 API, and how to upgrade previously done simuations.

The workshop can be done on-site and remotly. We will do our best for remotes attendees so they have the best experience. The sessions will be streamed and recorded and dedicated live visio-conference room will be created. For the “hands-on” sessions, additionnal member of the DEFROST team will be present online to support and guide remote participants in their modeling soft-robots.