Art and Robots

Uncategorized 4 Comments

kim and stephen inspect the robot
Kim and Stephen inspect the robot – it is a large orange arm that can hold almost any tool, and move with incredible precision exactly where you tell it to.

In which Sugar vs the Reef artist, and veteran sculptor Kim Williams recounts our first steps into the high tech world of manufacturing robots…

 

Today Lucas and I visited a large, high-ceilinged industrial scale workshop at the rear of Building 39. It is the site where our 3d model of the Mackay catchment is being fabricated.

Earlier this year we took up a serendipitous opportunity: Zengxi Pan, an Engineering and Robotics fellow, wanted to work with artists for a Global Challenges grant. At that time this we were already planning to build a 3-d model for the Sugar vs the Reef project that Lucas is driving. It was an opportunity too good to resist.

 

mackay map showing area for which want to make a 3d model
Mackay map showing the general area for which want to make a 3d model. The idea is to physically manifest the landscape on a smaller scale, so farmers and everyone else can get a sense of how the water (with sediment, fertilisers, pesticides etc) flows from land to sea.

 

At our first meeting with Zengxi (Stephen), he confidently predicted that the robots could do everything we wanted them to do and more. It was as if the robots could cancel out human effort entirely, once you gave them the correct commands. We went away from that first meeting amazed and amused that a problem that we had on our list to solve (but which we had no idea how to solve) was effectively being solved for us.

 

3d model digital simulation
3D digital simulation of the Mackay Pioneer River Catchment area. The data comes from an open source Digital Elevation Model.
dataset
What the data set actually looks like – a boring grid of numbers, each of which refers to a location point on the landscape, and that point’s corresponding elevation above sea level. We have learnt so much in these past couple of months!

 

Since then we have been liaising with Zengxi and his research fellow Donghong Ding, an affable young man who is doing all the hard work. My role as research assistant in this 3d project is to do some of the background work. I’ve sourced the data set that was needed for the model and have done materials research and of course met with Lucas, Stephen and Donghong on a fairly regular basis over the last two or three months.

 

dong hongs working out
Whiteboard in the Robotics workshop – these are Dong Hong’s sketches where he is trying to work out the most efficient method (in terms of time and materials) for making our 3D model out of plywood using a router.

 

I have now realised that Stephen’s confidence was broadly based. His claims about the ease with which the robot can perform the task of fabricating a 3-d topographic model have a few caveats. In fact it is a highly experimental process and Donghong is immersed in an intensive process of problem-solving.

 

donghong with the robot
Dong Hong with the big robot arm. In the foreground is his first experiment using a 1mm router bit, moving in tiny increments of 0.1mm for each pass.

 

The robot they are using is like a giant articulated arm. For our 3-d model, it is fitted with a router which is subtractively carving the model in plywood, one sheet at a time from sea level upwards. It is a painfully slow process, depending on the size of the router bit being used at any given time. A 10mm router bit can do the gross cutting away of material, to be replaced with a 1 or 2mm router bit for the fine detail. The bands of plywood create very beautiful topographic layers which highlight the elevations through colour variation:

 

trial 3d model
A small (15cm square) “proof of concept” model which Dong Hong and Stephen made before beginning the bigger version. It’s beautiful, and we really like how the layers in the plywood operate visually as elevation contour lines.

 

Donghong is grappling with the best way of configuring the robot to perform this task. He is doing this with industrial output in mind: how to use the least amount of materials and how to perform the task in the most time-efficient way. It is an incredibly slow process, as the robot moves the router back and forth across the plywood in tiny increments. To this point, using the method Donghong devised, the robot is still on the first sheet of plywood. Ultimately the work will require about six thicknesses of plywood sheeting to reach the required elevation.

 

the 3d model - base layer
This is just the base layer (several sheets of plywood will stack on top of this one) – but you can already see the beautiful estuarine shapes emerging.
work in progress
Work in progress on first iteration of the 3D model.

 

As the elevation increases, the robot will eventually be cutting out very small shapes which represent slices of landmasses. Donghong then has to match the pieces and glue them together by hand. This strikes me as a very painstaking task, but in Donghong’s view it will be the most efficient and economical method. I am offering him some woodworking expertise when it comes to gluing and clamping. My job now is to make some soft weights, which I’ll probably make with socks and sand or rice. These will act as clamps, laid gently over the complex small shapes while they are being glued to their base layer.

 

 

Of course it would have been easier to just glue the requisite number of plywood sheets together and carve subtractively from the top down, but this would apparently lead to a 90% wastage of material. Donghong is not happy with the current cutting program, so it’s back to the drawing board – he is devising a new method which should make the job faster, so he will start the process again from the beginning.

 

donghong changing the router bit

 

We asked if we could keep the test piece, as it has yielded fascinating marks and shapes and is a good example of an experiment-in-progress. Donghong was a bit bemused by our request – why would we want to keep a failure?

 

test model rejected