Soft Robot Success

And the winner is utter simplicity . . .

Three short segments of latex tubing without any attempt to glue them or bond them together except at the base and the top. Each tube can fill fully and independently and is not so long as to have just one portion blistering and the remainder uninflated. As can be seen in the short video while the air still causes a bubble almost the whole segment is involved and the robot can be moved easily as each tube is fully inflated.

The next step is to build a line of these short segments into a stack of linked robots. We expect that a stack of small bubbling segments will form a controllable length of robot.

Our first pass will likely have the pneumatic feeds brought out externally to the robotic bundle to test with as shown above, but with experience we plan to move the feeds into the core.

While our robotic assembly is simple, the control mechanism is getting complicated. To control the robotic bundle we have recently acquired a 24VDC solenoid valve assembly consisting of 20 independent air valves.

To control the valves, we need an electronic controller. We are using discrete FETs to drive the 24V solenoids which we can't directly interface to a low voltage microcontroller. While at one point we thought the design was done, we made the mistake of asking some experts for advice. While generally seeking the advise of people who know things better than yourself is a good idea, asking too many experts makes things confusing. As Frodo stated 'Go not to the elves for counsel, for they will say both no and yes' and in this matter we have learned that every expert provides different opinions on how to drive anything.

Some experts want a Zener diode to the low side, some want a silicon diode to the high side, others want a RC circuit you tune to the coil in the solenoid (and also the wire, connector and everything else between the drive and the air valve.) As a compromise, we've designed a board that incorporates both diode approaches but not the tuned RC. We have uploaded the revised board to OSHPARK's small run service to build the board, but while the board passes their integrity check, we're holding off ordering until we test the circuit on a perf board. Again, once the final board is populated and tested we will release all the board files as well as general documentation on why it was designed the way it was and how we suggest you use it. In the meantime, the revised schematic is available above; if you spot an error please drop us a line.

Then it is on to assemble the whole lot and see if we can grab a horse shoe! Hopefully this does not constitute getting cocky, but, the pieces of the controller of the robot stack are now the background for our updated website logo.

Much to do, come back often!

Special note: Thanks to Bill & Darlene at Atlantic Rubber Co (37 Ayer Road, Littleton, MA) who provided tubing to help the project. Not only a well stocked local source of material but also willing to help out a high school student's project.