When I was a student at the Slade I was working largely with abstracted form grammars as pictured above. Now that I am coming to the end of the MSc at the Bartlett, I’m thinking about how the Soft Robotic actuators that I have been working on could form some kind of addition to this, and in general how they could be integrated into creative practices elsewhere…
The main thing that I would hope to have come out of my work is that in looking at the problems regarding material tolerance and durability, I have suggested an alternative strategy that would enable a greater degree of ‘free play’ in designing the actuators. This has largely to do with scaling to some degree, but above that, an understanding of the aspect ratio of the internal network is key to success in my opinion. In lieu of a fully formed model of the behaviour (there isn’t one in existence) I think there is a place for soft robotics in the design context, but there are some fundamental principles that can really cause major problems, especially if they are to be integrated into a creative work flow. I hope that my work has gone some way towards overcoming these problems and will enable like minded thinkers to get involved in this field.
For more info: http://benhaworth.tumblr.com/
Pretty early on I decided to build myself a piece of parametric software to define the geometry
for the actuators. Actually I built two separate programs… The first was essentially the standard
parallel layout with the central channel and the second was based on Bezier curves for the
main component of the network.
When defining input parameters over multiple iterations… Having a program
set up proved incredibly useful in labour saving as I was having to draw all my geometry
manually previously which just got tedious. Also, it was great to have all of the input variables
stored automatically (a spreadsheet is exported with the geometry)… I had the feeling that
generating data based on the actual material performance would be as easy but hmmm…
I guess we live and learn!
…the software here was scripted in Processing
Connecting the BMP085
I managed to get the BMP085 Barometer connected fairly easily and its giving stable readings with a tolerable degree of noise. The image here with the Blue dots is atmospheric pressure readings in Wolfson House (UCL) that are being fed into processing Via the Arduino. I am getting a reading of approx 102158Pa (Pascals). meteoprog.co.uk is saying that the pressure in London today is 796 mmHg(Torr) which converts to exactly 102524.619 Pa - so I’m satisfied that the Bmp085 is giving me pretty reliable data.
The most useful literature I found about connecting the Barometer was provided by the good people at SparkFun here. There were two fixes in the code as I think some of the functions have been updated:
- wire.send(); becomes wire.write();
- wire.receive(); becomes wire.read();
I think I’ve got as far as I can with the Whitesides layout for the Pnuematic Networks. I don’t think that in its current form the material system is significantly scalable (or even durable at its current scale). I’ve gone through every variable to test which part of the system is causing the most problems.
I have narrowed it down to the central channel that is used to distribute the air throughout the network. Initially I had thought that the effect of this channel was fairly benign. However, the inflation of the channel as seen in this image, is causing a discontinuous stress distribution in the partitions between the network cavities.
The cavities are clearly inflating the most in the middle (most compliant part) of each. But In the above example, the weakest part of the partition is in the middle - at exactly the place where it should be the STRONGEST!!!!
In removing this channel, I am starting to see enhanced potential for scalability (see Below).
Ruairi Glynn and Stephen Gage at the Bartlett have been asking about the potential to scale
up the actuators for quite some time now. I’ve been ‘umming and ahhing’ about this for a while
and have thought of a list of reasons why it wouldn’t be possible. Recently I decided to bight
the bullet a give it a go. There’s been countless tests to get to this point but some subtle
tweaks in the mould material and construction of the moulds has made this possible.
In fact I think up scaling the material system is not only possible but actually essential to some
degree as it provides:
- Increased robustness
- Increased tolerance for human error during construction
- An enhanced seal when adhering the two parts (Compression and tension side)
- More possibility to make repairs if necessary
- Cheaper mould material due to changed methods
- Increased variety of potential applications