Rigid and ‘hard’ traditional robots are going to hurt if they hit you. This makes robot and human interaction tricky and we need to add sensors and use clever methods to avoid this. Soft robots are a new class of robot that are made of soft and spongy materials and use pneumatics to actuate and move. This opens a world of problems in human-robot interaction that soft robots can help solve!
Another thing soft robots are good at is adaption. The soft materials can conform around the environment, survive falls, and squeeze into tight spots. Because of these characteristics, soft robots have used biology as a major inspiration. We’ve seen soft robots inspired by an elephant’s trunk, caterpillars, worms, maggots, octopuses, fish, lamprey, spiders and a whole bunch of others.
The field of soft robotics is very young, an infant in the grand scheme of things you might say. We can make some cool soft robots and make them do some interesting things, but most soft robots are not that useful yet. I believe that soft robots are not useful yet because of the way they are made, and how we make them. I think that we are limiting how useful they can be.
We wrote a paper for a special issue on Soft Robotics in Biomimetics. We looked through the literature and noticed a trend in the design of soft robots. Say we think up some new type of soft actuator and then we go off and design and optimised it. Let’s call this a ‘functional block’. We go to our labs and try to make our functional block do something interesting and novel. The trend in the literature is to combine or ‘stack’ many functional blocks, and this creates more capable systems with more complex behaviours. A single functional block might only move up and down, while stacking several functional blocks can enable a range of complex and diverse motions with increasing capabilities. Functional blocks that are stacked create systems that are greater than the sum of their parts. This is a classic engineering approach.
But we noticed a downside. The prevalent model for the control of soft robots is to have one controller output for every functional block. Softs robot will quickly run into practical limits on size, capacity, and energy requirements. If we need 100 functional blocks to imitate an octopus’ arm, then we need 100 controller outputs. This is well beyond any practical limits in control and this confines the capabilities and usefulness of soft robots.
So, we have made a prediction. As we move towards more capability in soft robots we will require increased amounts of autonomy. We will see more functional blocks with fewer outputs from controllers to deal with the complexity of controlling these stacked systems. Time will tell if this prediction comes true.
Recent Comments