Author: johnmarshall


The tea-house is heading to Kyoto, Japan.

The ‘skin’ rolled up as one piece.

All the other components were braced together…

… and a crate built around them.


A series of tests of the tea-house ‘eyes’. These are done with passive infrared sensors.


Another (almost sleepless) week. Some of the many things that happened:

The tea-house structure was assembled and painted.

The tea-house skin was laser cut and ‘stitched’ together. The skin is made from 2 layers of precisely cut synthetic paper pieces that interlock with each other.

The robot electronics and coding were 90% completed (still some tweaks necessary). We are grateful to Osman Khan for all the time, patience and knowledge that he brought to this part of the project.

The operable panels (a.k.a. the tea-house ‘eyes’ were mounted and strung).

Half of the tea-house skin was test hung.

Chris Johnson fixing the skin.

(Westley) Josiah Burger inspecting the skin seams.

While the skin was in place we did a test of where the robots will be seen and finalized the design and fabrication of the platforms they will inhabit.

A detail of the tea-house skin.

TST_003 and RDO_002 look through the tea-house ‘eyes’.

When the ‘eyes’ are activated they allow for line-of-sight between the robots and any human beings that might be interrupting them.

More work was done on the tea-house skin and the entire surface was test hung.

A lighting test was carried out.

Cezanne checks the view through a tea-house ‘eye’.

We tested various ways of lighting the tea-house.

RDO_002 ready for paint.

TST_003 ready for paint.

MXR_011 ready for paint.

TST_003 during its ‘toast’ cycle.

Tests of the tea-house ‘eyes’:


A busy week. Here is some of what happened.

I modeled the front of TST_003 from the dimensions of the cardboard mock-up. This incorporates locations for the sensors and guides for the toast ‘drawers’. The parts were run on the FDM.

A piece of ABS sheet was cut out and fitted into the top half of the shell of RDO_002. The same was done for MXR_011. These are to locate and hold both pieces of these shells in place.

These were fixed in place with ABS /Acetone slurry. (Note the sanded FDM part of TST_003  in place in the image above and in the video below).

An ultra light density fiberboard (ULDF) buck was cut, sanded, sealed and polished to be vacuum formed over for the rear end of TST_003.

‘Nubbins’, dials, drawers, handles and sensor holes were made, cut and inserted.

Side panels for TST_003 were cut from ABS sheet and joined to the FDM front and vacuum-formed rear part.

Cezanne began making the circuit boards for the final versions of the electronics.

Robots ready for final integration.

In the meantime, Chris Johnson has been working on the tea-house structure and Westley Josiah Burger has been working on the tea-house skin.


I built a functioning prototype of the TST_003 toast mechanism (using the L12 linear servo) to integrate with the chassis and the robot shell. The photo below shows a comparison with its inspiration (the rear end is an Airstream trailer).

Now that I have something to take actual dimensions from, I can model the mid section of the robot shell for fabrication.


An update from Karl, Chris and Westley on the tea house panels. The ‘eyes’ will open and close to reveal the robots inside based on Omron’s Okao Smile Scan (see below).


The ‘speaker grill’ of RDO_002 could not be included in the vacuum formed shell since it is a series of undercuts. Therefore a separate piece was built on the FDM and needed to be ‘scabbed in’ afterward. We could just have glued it in place and filled any gaps with body filler but this tends to shrink and crack with age and abuse. So I used my ‘dissolved ABS shavings’ method (mentioned previously).

The part to be inserted was traced around and cut out. I do straight sections with a scalpel and whittle away at curves with a rotary cutting tool. The edges of the hole are cleaned up with a (high quality) file.

The part to be inserted was held in place (on the show surface) with masking tape and both parts were fixed in place on a sheet of Aluminum.

I made some shavings from the off cut of ABS plastic (with a deburring tool – the thing with a red handle in the picture) and dissolved them in Acetone (in a glass, ex-apple juice bottle). The Acetone/ABS slurry was painted onto the joint making sure to overlap both pieces. It bites right into both parts and makes the whole thing as one. The melted area stays ‘chewy’ for a while – so it was left overnight to set up.

The parts were now fused but the joint line was still visible with a small gap.

Most of the show surface was covered with masking tape leaving a small exposed area either side of the joint line. This was painted with more Acetone/ABS slurry. As the slurry was getting ‘chewy’ it was trimmed and the masking tape removed to prevent it getting stuck under the fused ABS.

Again it was left overnight to set up. After some sanding, RDO_002 now has a ‘speaker grill’ that won’t shrink or crack or pop out if too much flex is applied to the shell.

I also made the handle for RDO_002 while waiting for Acetone/ABS slurry to set up.

TST_003 now has a linear servo for the toast cycle and the chassis was adjusted to be lighter.


The gubbins have been fitted to the robot shells. Some cosmetic work has been done and  progress has been made on integrating the mechanical and electronic components.

Shawn O’Grady in the UM3D Lab built some parts for RDO_002 on the FDM. Since the build envelope on this machine is so small the ‘speaker grill’ had to be built in two halves. However, I took some shavings from an off cut of ABS plastic and dissolved them in Acetone. This makes a great glue. I had used this technique with ABS sheet stock before, but not with an FDM part. It worked out even better than I expected – the Acetone/ABS slurry wicked right into the part and the whole thing is one piece, now.

One of the big challenges left to overcome is the specific mechanism for the toaster cycle of TST_003. In the picture below you can see the toasting ‘drawers’ mocked up in cardboard and masking tape to see what potential interferences there might be.

Testing MXR_011’s ‘spin cycle’.

RDO_002 is too cute for its own good.

Using a servo and a gear to prototype the toasting mechanism of TST_003 was less than encouraging.

As an alternative, I’ve ordered a linear actuator. If it works this should free up some real estate and get rid of some additional weight.

In the THR_33 installation, the robots will ‘live’ on counter tops within the tea-house structure. We need a way to keep the robots from driving off the edges. Osman Khan accelerated the development of this  by writing some code for some QTI sensors so the robots will follow a line.


The robot shells have been trimmed and a second set was vacuum formed (just in case).




Now begins the process of structural, mechanical and electrical integration with the shells of the robots.