Robot Sculptor: Robot Cutting Clay for Mold Making

March 19, 2012
by admin

For the past few years I have walked by the KUKA robot at the University of Michigan architecture FabLab and couldnt wait to use it.  It took until this Masters of Science program that I was able to do work with the robot.  During my BFA in sculpture I had a few ideas for what I wanted to do on the robot.  Many of the projects I would like to do are translating human sculptor techniques to robotics.  Lately I have been working with paintings as three-dimensional objects, making me think of using the robot to imitate a painters strokes.  Another idea stemmed from the work Marin Abell did back in November of 2011 with motion capturing a sculpture chiseling out a form.  Being able to record the movements of a traditional sculpture activity and recreate them with robotics opens new doors to the way we look at art.  So I pose the question: If I could sculpt something from marble and record my actions, then have a robot cut out that same object in the same manor I did, do we lose or gain value in the recreated object?  At first I think we can say it is devalued, it is not the original artists hand work, but during this time of art and technology, I think the robotic recreation would be put up on a pedestal as a work of art just within the process itself.  Maybe this is just the talking dog syndrome, but there could be more to it.  So I wonder, how exact can I mimic the work of an artist.  If I motion capture the artist and track the tool, what information can I get?  Well, the path in which the tool takes can be recorded, the body angles of the artist can be recorded, and since the recording is in the time domain, velocity can be recorded.  Now given the artists weight we can compute the forward dynamics and start to understand the material resistance.  So in the end, I think we can get pretty close, but one missing element I have seen with the with all these processes is the tool in which the robot uses.

The tool the robot uses bring up many questions.  The basic question is, which tool is the “best” for the job.  The decision of what is best becomes a combination of expense, time, and accuracy.  This is in relationship to the material to be cut, and in the case of something like wood, the most common option is the router bit.  This brings me to the convergence between a traditional art background and digital fabrication within architecture.  As a sculptor, I use clay in more that 75% of my work.  Clay allows for an extremely malleable material than can be adjusted for rigidity.  So when I use clay, as well as the tools, I think to myself, there is not way a router bit could create this.  There are various sizes of router bits, and with infinite time and money a small enough bit could be used, but it still would be unable to accomplish certain things.  Within a more realistic approach, trying to recreate something done with a clay tool with a router bit would leave a different texture/pattern.  This requires post-work to resolve, meaning more hand labor.  This texture becomes an issue of not having the right process, so more work needs to be done to resolve it.  Another issue is with undercuts.  With multiple axis robots we can begin to create some undercuts, but mostly at a large scale.  All of this culminates to the use of the same tools a sculptor uses.  As Sculptor and Professor Louis Marinaro once told me, why change the tools if it worked for hundreds of years.  This philosophy made me question, why doesn’t the robot, with so many axis it can imitate my arm, use the same tools I do.

At the beginning of the fall semester our MSDT program was split into two groups for research.  A week of conversation as to who wanted to do what led to me working with Jason Prasad at Electronic Delectables and Sonia Tereszczenko at Digital Undertaking.  With Jason from Engineering, Sonia from Architecture, and Me from Art, we formed a true multidisciplinary team.  After discussing our interests and proposing ideas we decided on creating sculpture tools for the robot.  While our initial plan was to start with clay as a practice run for other tools, we have worked our way deeper into the clay research.  While I have previously wanted to do clay sculpting with the robot, it was not until our group conversation with our professor Wes Mcgee, that my thoughts on clay sculpting with the robot and being able to make molds of it, became making final products on the clay.  Right away I thought of fiberglass as a method for final production.  This was probably because two years ago I used clay as a base material on a motorcycle panel and created fiberglass shells off of it.  Unfortunately, that project was terminated due to me losing the motorcycle 🙁 .  Before I get any further into the project, I will leave you with some photos of the test run, which worked pretty well.  After making our own clay, we constructed a box to press the clay into, water jet cut some tools mimicking the ones from sculpture, blocked out an area on the vacuum, and created some toolpaths using MasterCam and the Robot plugin.

I have to say, I am extremely happy to be working with my group members, and I think this project in itself shows the power of collaboration and different disciplines coming together.  Our advisor Karl Daubmann brought Jason and I into the program after knowing us from his (and John Marshall, and Max Shtein) multidisciplinary class Smart Surfaces.  I hope in the future there are more programs that allow this type of culmination of disciplines to come together and learn and research together.  Unfortunately the architecture school at Michigan does not seem to be on that same page, announcing that they will only be letting architects into their program, so it may be an end to these types of projects. 

Here is our first test.  We made this as a proof of concept.  The stages were from packing the clay, putting it on the vacuum table, cutting paths out of the clay with the robot, and then doing a fiberglass layup.  Some issues need to be resolved, like the best way to clean the clay off of the fiberglass, as an 1/8″ or so sticks (but we may be able to resolve this with the proper mold release spray) .

Kuka Robot Cut Clay

Kuka Robot Cut Clay

Kuka Robot Cut Clay

Kuka Robot Cut Clay

Kuka Robot Cut Clay
Kuka Robot Cut Clay

Kuka Robot Cut Clay
Kuka Robot Cut Clay

Leave a Comment