The future of actuation is soft and EFFICIENT

Emptying the dishwasher or hanging up laundry, who hasn’t wished for a robot which could do these necessary but time-consuming and annoying tasks? I certainly have. Humanoid robots today are computationally very intelligent; however, they rely on bulky electromagnetic motors that cannot replicate the physical intelligence, which living creatures have through their soft muscles, tendons, and ligaments. The lack of physical intelligence makes these robots less adaptable to unforeseen challenges in the home. To add to that, many tasks require strong but slow motions (high torque and low rpm), which are inefficient for current electromagnetic motors and causes them to heat up rapidly. 

Replacing the electromagnetic motors with soft components, in particular soft artificial muscles, could facilitate these tricky household tasks, and much more. There are many types of soft artificial muscles, and our research group focuses on soft electrostatic actuators which, much like the electrical impulses that trigger natural muscle, are driven by electrical signals, allowing fast, powerful, and silent actuation. In contrast to motors, soft electrostatic actuators do not require any energy to hold a position,  and thus, we hypothesized, would be more energy-efficient for typical robotics tasks. 

Energy efficiency of the actuators is very important because it determines how much energy robots need to move, and thus how long they could run around with a battery pack. Having a robot run out of battery before it finishes its rescue mission is not ideal! Unlike motors and engines, for which efficiency has long been thoroughly studied, the efficiency of soft actuators has been barely studied at all.

The goal of our project, led by Steven L. Zhang and supervised by Christoph Keplinger, was therefore to understand what governs the efficiency of soft electrostatic actuators and then to improve it. The first step of this project was to decide what method to evaluate efficiency to use, because there is no consensus in the field. Of course, efficiency is the ratio of output to input energy, but for what task? Moving a dumbbell up and down, pushing a box across a room, or maybe lifting a mug from the kitchen counter up onto a shelf and then coming back down empty, ready to grab the next mug? We insist that the last example is an appropriate task because in that case, work is done and it is a closed mechanical work cycle. Closed cycles include all energy flows, including the work required to bring an actuator back to its original state. To methodically evaluate closed-cycle efficiency, we next had to develop an appropriate experimental setup. 

We chose the HASEL artificial muscle, a promising class of soft electrostatic actuators, to study efficiency. We put the actuator through its paces, testing it across a broad range of experimental conditions, such as force, voltage and frequency. We discovered that with the right conditions, HASEL actuators are much more efficient than previously reported – 63% instead of 21%! To show that our method is applicable to other technologies, we also tested a different type of actuator, a dielectric elastomer actuator, a more established technology in soft robotics, and similarly uncovered, that the right combination of experimental parameters results in a higher efficiency.  

Ultimately, we hope the framework introduced in this paper does not just provide new knowledge, it is also a reminder of the unlocked potential of soft electrostatic actuators. While researchers have spent many decades refining electromagnetic motors, soft electrostatic actuators are a young technology with much left to be uncovered. By discovering that soft electrostatic actuators, such as HASEL actuators, are far more efficient than previously believed, we are unlocking new opportunities for realizing highly adaptable, multifunctional, and energy-efficient robots that will help us at home. 

Publication: 
A method to study and enhance the energy efficiency of soft electrostatic actuators
Steven L. Zhang, Toshihiko Fukushima, Sophie Kirkman, Soo Jin Adrian Koh, Philipp Rothemund, Christoph Keplinger
Proceedings of the National Academy of Sciences
https://doi.org/10.1073/pnas.2527676123