Soft Robot Failures

We learn from our failures, at least we hope we do. Below are a couple of experiments we undertook that we want to share.

Gelatin Body Robots

Silicone may be the production material of those with good facilities, but gelatin is the wobbly stuff found in the average kitchen. Since our lab is our kitchen, gelatin seemed a natural choice. Gelatin however just proved too fragile to withstand even low pressure pneumatics.

The basic construction techniques were the same for all our attempts:

  1. My Science Project contains a recipe for the gelatin we used. It is simple and uses commonly available grocery store dry gelatin.

  2. Twisty Latex Balloon Animal Balloons were our air bladders. We settled on these because they were thin and long like the structures in the SEAS tentacles. Ours were from a local party store.

  3. Red "Solo Cup" was our mold. Again, readily kitchen available and they seemed appropriate for the job at hand.

  4. 1/4 inch flexible plastic tubing was used to connect to the compressed air. Readily available at our local hardware.

The basic construction method was the same, attach balloons to the tubes, pour gelatin around the balloons while they were in a cup, cut the cup off the gelatin after it had set. We made three passes at Gelatin Body Robots (GBRs) before moving on.

GBR1 - the first attempt at a gelatin robot was to try to drop empty balloons into a cup of gelatin. However, BALLOONS ARE NEVER COMPLETELY EMPTY and so they were buoyant in the gelatin. They floated to the top and we never got to the stage of applying air to the balloons. Not just failure to launch, failure to reach the launch pad! If you look closely, you can see the pink balloon starting to curl in this photo.

GBR2 - After the utter failure at our first attempt, we modified the built process to insert very thin dowels into the balloons to ensure they would submerge. It sounds simple enough but with the tubing attached to the balloons prior to submersion in the gelatin we needed long and unwieldy dowels to make thing work. Unfortunately, we did a poor job of air bladder placement and they were so near the gelatin surface that the balloons immediately broke through the gelatin and did not deform it. Click the picture above to witness the quick death of our second attempt.

GBR3 - Knowing that the gelatin was fragile, our third attempt moved the balloons well inside the body of the robot. We used shorter tube segments so the dowels were not quite as unwieldy and let the gelatin set for over a week to ensure it was fully congealed. As you can see in the video if you click the picture, it was better but it still did not work. While we were able to get one side of the material to deform, we never saw the movement of the material as a whole in any direct . . . and then of course the side ruptures.

In summary - Gelatin lacks enough strength to hold together even under slight pressures. When we were able to inflate the bladder in GBR3, we saw the side expand out but the body of the material never rotates. So we are abandoning gelatin as the material for the body of the robot.

There is one hope for gelatin, we are trying to research ways to make bad gelatin. All of the recipes we have found so far tell you how to make gelatin that is, well, gelatin. We have recently tried to reach out to people who may know about gelatin for ways to make what may be bad food but good robot material. But, as we said before, we're off of gelatin unless we find another way to prepare it.

Silicone Laminated Latex Hoses

Looking around for ideas, we settled on using latex hose as a bladder and then coating it with silicone. The thought process was that when we tried to use just silicone in other projects, we were unable to avoid bubbles and gaps in the material which would make it completely unsuitable for inflation with air. If we started with an existing bladder then covered it with silicone we would not have issues. Rubber bands were added at regular intervals to help ensure the three tubes would behave as a single assembly.

However that was not to be. As can be scene in this video, air pressure starts a "blister" in one small area of the tube and then the blister in just that one area grows. We could never get a smooth curve or continuous bubble region in the laminated structure. We added huge numbers of rubber bands and tried other constraints but the result was always the same, one huge bubble and no control of the expansion. The bubbling area would push the constricting rubber bands up or down the length of the assembly regardless of how many rubber bands we applied or how long we had let the silicone set.