would fabricate the 243 geometrically distinct plates
of beech plywood that comprise the 245-square-
meter ( 2,637-square-foot) structural shell, which
resembles the shell of a peanut.
Before the robot could do that precise work,
however, the project team ;rst had to program it
so that it knew how. ;e team also had to teach the
robot how to work with the other digital machines
involved in the build.
“Our challenge was to make the machines intelligent,” says Mr. Schwinn, who served as the lead
project manager. “We were interested in geometric
di;erentiation designed for individual requirements,
not mass production. Every piece had to be unique.”
To create the software program that would
design the interlocking plates, the project team
studied and emulated the microscopic connections in the plate of a sand dollar—thus adopting a
biomimetic approach. It took the project team six
months to develop the technology to de;ne each
plate’s geometry and generate the code that enables
the robot to manufacture each unique plate.
“We had to create a unique robot program for
each piece,” Mr. Schwinn says. “To make the pro-
cess e;cient before we moved to fabrication, we
developed the tools that would generate the robot
code for every piece beforehand.”
Once all the tools were developed, a human oper-
ator needed only about one minute to generate the
code for each plate; the robot took about 20 minutes
to fabricate one plate. ;e robot completed all the
plywood plates in just over two weeks, and the exte-
rior shell was constructed in just four weeks.
;e robotic design yielded remarkable resource
e;ciency: With a load-bearing structure that’s only
50 millimeters ( 2 inches) thick, the pavilion needed
just 12 cubic meters (424 cubic feet) of timber. Only
about a tenth of the project’s €425,000 budget went
toward the shell’s materials.
“A building like this would be impossible without
the use of robotic fabrication and digital processes,”
Mr. Schwinn says. “;e robot’s kinematic ;exibility
is a requirement for the production of such complex
and individual geometries.”
;at’s precisely what makes robotic technol-
ogy so promising for timber construction: No two
structures are alike.
“We can design buildings according to site-and material-speci;c requirements. We don’t have
to rely on industry averages,” Mr. Schwinn says.
“;ere are no cookie-cutter projects.”
;e Landesgartenschau Exhibition Hall—the ;rst
structure entirely made of robotically fabricated
beech plywood plates—opened in mid-2014.
“Machines are
absolutely
capable of
doing this type
of work. The
hard part is
programming
them to do so.”
—Tobias Schwinn, University
of Stuttgart’s Institute for
Computational Design,
Stuttgart, Germany