Why soft robots ?

Not the most comfortable attire. Made using the midjourney AI.

What's the point of soft robotics anyways? 

Creating robots using softer and lighter materials fundamentally change where and how to use these robots. Soft robots should not compete with traditional robots but rather complement them so that automation can reach new applications and hopefully improve human standards of living. There are several applications in which soft robots are expected to outperform traditional rigid robots:

Human interactions. Why do we wear everyday clothes when we could wear plate armor ? The answer is that, unless you are doing historical European martial arts, you don't expect to be bashed with a morning start anytime soon. Thus, it makes no sense to prioritize plate armor's durability compared to regular fabric's comfort and low weight. Although this admittedly a far-fetched example, if you were to be carried by a robot, of if you were to equip some sort of an exoskeleton, the softness of the robot becomes an important aspect. In general, if you have close physical interactions with the robot it may be interesting to use soft robots. 

Adaptability. Traditional industrial robots are usually measured in terms of speed, reliability, and force. You want them to do a single job all day, every day, and without interruption. Soft robots on the contrary are more suited to operate in uncertain environments. Indeed, one of the advantages of soft and lightweight structures is that they can elastically deform instead of breaking, thus being ideally much more resilient. Another advantage of soft robots is that they may be able to navigate cluttered and/or unknown environments, or conversely, grab object of unknown or changing sizes. This is due to the intrinsic capability of soft structure to deform to fit and conform to their environment.  

What are the challenges and limitations of soft robotics?

Control & kinematics. In robotics, control refers to the ability to predict the position and shape of a robot given a specific set of commands given to its actuators. For example, in a traditional robot, let's say you want to move your robot to grasp a specific object. First you need to determine what positions the robot needs to move through to end up at the right place, and second you need to know which commands to give to your motors to in this way. All these challenges are well known and well handled in traditional robotics and form the basis of traditional robotics.

In soft robotics however, our robots are soft and our actuator technologies are less mature. The first issue we may encounter is to try to move a soft robot in a predictable fashion. After all, if your robot is really soft in all directions, how do you reliably know in which direction it is going to deform ? The second issue is related the actuators used in our soft robots. Soft actuators in general both move in a less predictable fashion and are harder to measure as well. Both these issues make it challenging to control complex soft robots. 

Force generation. For soft robots to be useful in some ways, they will need to somewhat move and push things. However, when your whole body is soft, how do you generate forces ? Octopi have a body that is almost devoid of bones, yet they are surprisingly strong and capable of manipulation.  Elephant trunks are also boneless, yet also strong and agile. However, these are arguably exceptions. Most animals, and especially terrestrials animals, rely on a skeleton to support the forces generated by their (soft) muscles. 

What does it say about soft robotics  then ? Well, rather than to consider the field of soft robotics as a separate domain, it's much more useful to consider robots as belonging to a spectrum, with rigid industrial robots at one extreme, and soft robotics blobs on the other end. In the future, it is likely that different soft robots will be more or less "soft", but even within a single soft robot, we are likely to see various materials with different harnesses. 

Don't we already have robots?

Yes, we do. But as explained above, soft robots won't replace traditional robots. Applications in which soft robotics is useful are new applications in which traditional robots are not used at the moment. 

References for further reading: 

[1] Hawkes, Elliot W., Carmel Majidi, and Michael T. Tolley. “Hard Questions for Soft Robotics.” Science Robotics 6, no. 53 (April 28, 2021): eabg6049. https://doi.org/10.1126/scirobotics.abg6049.

[2] Laschi, Cecilia, Barbara Mazzolai, and Matteo Cianchetti. “Soft Robotics: Technologies and Systems Pushing the Boundaries of Robot Abilities.” Science Robotics 1, no. 1 (December 6, 2016): eaah3690. https://doi.org/10.1126/scirobotics.aah3690.

[3] Majidi, Carmel. “Soft Robotics: A Perspective—Current Trends and Prospects for the Future.” Soft Robotics 1, no. 1 (March 2014): 5–11. https://doi.org/10.1089/soro.2013.0001.

[4] Pfeifer, Rolf, Max Lungarella, and Fumiya Iida. “The Challenges Ahead for Bio-Inspired ‘soft’ Robotics.” Communications of the ACM 55, no. 11 (November 2012): 76–87. https://doi.org/10.1145/2366316.2366335.