As the development of the artbot continues, many new and unexpected challenges are presenting themselves. One of these challenges is the inaccuracy in the (3D printed) artbot chassis. This means that the device needs significant callibration. For example:
- The wheels may be of slightly different diameters, so they will travel different distances
- The wheels may be different distances to the centre, so when the artbot pivots the pen will not be in the centre
- The wheels may be slightly warped (due to heat process of 3D printer) so when the artbot pivots there wil be different errors according to the current rotation of the wheels
- The pen holding mechanism needs to apply the right amount of pressure *and* stay within the centre
In the below image, for example, each of the leaves of the pattern should be perfectly centred. The variation is entirely a result of the inaccuracies of the hardware chassis. Of course, these inaccuracies are not necessarily aesthetically objectionable.
Perhaps the most interesting challenge with the artbot, however, is how to best control it. The current sketch (which is available on Github) uses a very simply designed API which essentially offers 2 instructions:
- Move in a straight line
- Rotate on the spot
This API results in a great deal of time lost when the pen is not drawing … the artbot spends close to half its powered time just rotating (so that it can draw another line in a different direction).
The question is: what other control mechanism might be more in-tune with the underlying structure of the artbot: a pen that sits in the middle of the axis between 2 wheels.
If one considers that drawing a curved line just requires each wheel to rotate at a different speed, then it becomes clear that it is almost easier for artbot to *think* in terms of curved lines.
This approach will be the next approach to exploring artbot.