Schedule

Relevant metrics for soft actuators include: strain, blocked force, speed, hysteresis, cycles to failure, efficiency, energy density, power density, force density, and mass of the power supply or support equipment. As a community, we report the best numbers for all metrics, but often neglect to point out that not all can be achieved simultaneously. This understanding is however needed to choose the actuator best suited for a given application. We tend to insufficiently acknowledge the inevitable trade-offs, such as between lifetime and strain, or between force and displacement. Our soft devices can be far lighter than the power supplies or compressors we use. For practical devices it is important to include power supplies in energy density calculations, yet we often omit them. Our math is correct, but are we giving the full picture? Using examples of DEAs and zipping electrostatic devices from my lab, I will illustrate how one data set can be interpreted in different ways. I end with some suggested guidelines for more uniform reporting of performance metrics.

There has been an explosion of robotic hands and grippers with novel soft articulating appendaged; these are often driven by pneumatic/hydraulic channels or tendons due to flexibility and form factor. However, proximal to the soft hand, we often find more traditional pumps and motors to articulate these transmissions. In my talk I will address the actuator selection and control decisions that emerged during the development of (1) a tendon-driven multi-finger hands for marine operation and (2) flow control of a soft suction gripper. Considerations reach far beyond stroke and strength, but include weight, resilience, and mechanical integration.

Soft actuators have a wide range of beneficial behaviors and potential downsides when they are applied to create desired soft robot movements. However, during design we often focus primarily on the intended motion and forces of the actuator and do not always consider all the additional factors that may affect the design suitability. In my talk, I will go through two examples of actuator challenges to make an argument for considering more factors during design. Specifically, I will show how characterization of a one-degree of freedom actuator using quasi-static shear and extension tests improves estimates of multi-degree of freedom parallel robot designs. Then I will discuss the design space of pneumatic actuators for shape control of soft growing continuum robots and how the design space can allow general actuation control.

Our coordination and convergence challenges in industrial soft robotics are defined by task uniqueness (no one has ever tried to do what we do) and expansive multidisciplinarity (...→chemistry→materials→mechanics→design→...). Such an environment makes it hard to scope and align research for the individual scientists, engineers, and designers, and makes prioritizing the work ambiguous for the academic and industrial organization.

The history of scientific discovery highlights certain research teams (Bell Labs, PARC, Rad Lab) that implemented strategic organizational and tactical behavioral innovations that lead to a culture of convergence. All research may be considered on a two dimensional space defined by the axes of Fundamental↔Applied and Curiosity-Driven↔Use-Inspired. Industrial soft roboticists are working in the quadrant of Fundamental, Use-Inspired research — called Pasteur’s Quadrant — that is neither pure basic nor pure applied research. My personal experience for the past four years has to been to observe the operation of this quadrant in Facebook (now Meta) Reality Labs Research with a focus on soft haptics.

Much like academia, industrial research into soft robotics is based on the fundamental questions that push each subsystem forward. The anatomy of soft robots spans actuators, sensors, control, power, and the mechanical structural (skin, bone, etc). The divergence between academia and industry is in terms of value assigned to system integration and to reliability and replicability. System integration is often considered merely a technical implementation challenge and without fundamental intellectual merit, and yet the gaps in human knowledge around holistic system design is vast -- especially in soft robotics. Risk assessment, balancing, and mitigation drive much of the industrial research portfolio management -- and a major attribute of this process is assessing the reliability and replicability of novel technology. The current academic incentive for novelty and first-to-press discounts the value of replicability and reliability that drives industrial research. Strengthening the alignment and bridging the divergence between academy-industry must be our goal in soft robotics research if we are to translate its impact more broadly to society.

Nature uses soft materials such as muscle and skin to build organisms that drastically outperform robots in terms of agility, dexterity, and adaptability. Biological muscle in particular is a masterpiece of evolution, as it powers the versatile arms of an octopus, is strong enough to move an elephant and fast enough for the wings of a hummingbird, self-heals after damage, and is seamlessly integrated with distributed sensing. These astonishing capabilities have inspired the creation of artificial muscles, a grand challenge of science and engineering that dates back all the way to the 17th century when Robert Hooke first recorded the idea.

The Robotic Materials Department at the Max Planck Institute for Intelligent Systems currently aims to understand the fundamental principles and materials science, as well as to develop robotics applications of a new class of self-sensing, high-performance artificial muscles, termed hydraulically amplified self-healing electrostatic (HASEL) actuators, a platform technology pioneered by this research group. HASEL artificial muscles achieve high speed and efficiency by harnessing an electrohydraulic mechanism, where electrostatic Maxwell stress activates soft, hydraulic structures to achieve a wide variety of actuation modes. Experiments show that current HASELs match or exceed most performance metrics of biological muscle; modeling results reveal rich underlying materials science to be further explored, and they lay out a roadmap showing how to drastically improve performance of HASELs, far surpassing both biological muscle and traditional electromagnetic motors.

This talk gives an overview of selected research results, including design of high-speed soft robotic devices powered by portable electronics; in particular, this talk will include discussion of experiments and methods to characterize soft actuator performance.

All robots rely on the transduction of energy from a storage form to mechanical energy. Soft robots, however, are far more sophisticated than conventional rigid robots in the transduction and use of energy to interact closely with the environment. They can exploit forward transduction to drive actuators, reverse transduction to deliver sensing and energy harvesting, and two-way transduction for in-material computation. In this talk we will use our recent research to examine how new metrics and test methods are needed in these different transduction mechanisms in order to deliver better soft robots, and how the lack of specifications in soft robotic actuators is holding back their development. Noting how closely and fundamentally soft robotics interact with their environment, we will also show how soft robot developers need to also consider high level metrics and assessments including safety, trustworthiness and ethics.