For two years I have followed soft robotics research including reading online papers, keeping up to date with The Soft Robotics Toolkit, and browsing YouTube channels. Most of my observations and inferences came from analyzing the information therein.

While soft robotic grabbers are renowned for their versatility and delicacy and have many potential uses in industry, soft robotic movers currently appear to be too slow and limited for practical use outside of the lab.

The most common methods of soft robot locomotion can be generally divided into three motions: walking, crawling, and rolling. Walking soft robots usually quadrupedal and move by rhythmically inflating and deflating each leg. While the specific capability depends on the individual robot, this entire group has some common weaknesses.  The first is a small stride. Quadrupeds generally move forward a very small percentage of their overall length per step.  In addition, all four legs have to take a step in order for the robot to move forward. This means that the robots need roughly 8 movements (4 if adjacent legs are moved simultaneously) for the entire robot to move. Larger robots should be able to move faster by taking larger steps; however, there are drawbacks. Walking robots have to carry their own weight; they are therefore vulnerable to the square-cube law. When a walker is scaled by a factor of x, its volume increases by a factor of 3, whereas the cross section of the legs supporting the robot will only increase by a factor of 2. This means that a walker that is simply scaled up will have increased difficulty supporting its own weight. Designers can support these robots by using stiffer materials. This introduces the conundrum that in order to enlarge the soft robot to increase walking speed, the robot must become less soft.

The second form of locomotion is crawling. I generalize “crawling” to be any locomotive task where an inactive section of the robot is dragged or pushed forward by an active section that generates motion.  This form of locomotion is much more simple and efficient than walking. Often robots only need two to four movements per step and can drag themselves forward up to 25% - 50%  of the robot’s length per step. Additionally, crawling robots tend to be easier to scale because in most cases the length of the robot determines the stride. Crawlers generally do not have to carry their own weight, leaving them less affected by the square-cube law.  In my observations, the largest weakness of the crawling robots is that both ends are dragged on the ground as the robot moves. Since neither end leaves the ground, the robots most likely have difficulty crawling over obstacles in their path.

Both the walking and crawling robots that I have seen either have not displayed the ability to turn or displayed considerable difficulties turning.

My goal is to create a soft robot that uses the best features of a crawler while developing a mechanism that allows it to climb over obstacles.