Robots have come to architecture to model, construct, fabricate and offer new approaches to create innovative designs, elements and structures

Industrial robots have been in action for over half a century. Applications are varied and well known. Much more recently, researchers in the architectural field have started to study how to apply the robot technology to advancing innovation in the area of architectural design and implementation. Architectural schools at universities around the world have been addressing these ideas for a number of years. To further the activity and to link researchers, the Association for Robots in Architecture was founded in 2010.

Much of the focus has been on how to implement the creative ideas that architects have come up with that traditional implementation methods did not address. Robots are addressing tasks such as fabricating structures using minimal materials, creating visual elements, building models quickly of complex structures and construction of actual buildings for labor saving and safety.

The research is mainly directed toward applying standard off-the-shelf robots with innovation in related areas such as programming software, designing custom robots, special-purpose end-of-arm devices and adaptation well beyond traditional applications.

In the software area, many of the developments are adapting software packages such as Grasshopper, Rhino and YOUR. Grasshopper is a visual programming language used in conjunction with Rhino. It allows the programmer to drag and drop components to construct three-dimensional (3-D) model design. Rhino (Rhinoceros 3-D) is a graphic modeling tool software package widely used in architectural robotics projects. YOUR is a software package used to create a 3-D interface from the computer. Together, these three software tools are used by many of the research teams.

In the end-of-arm tooling area, developments have addressed many new application areas, sculpting, metal forming and creation of innovative architectural elements. One research team has even addressed sculpting pottery materials.

Model building

One robot architecture driving force has been the interest in building very high-rise buildings in Southeast Asia. Some of the tallest buildings in the world are now present in Asia. Further driving the architectural interest in robotics is the fact that many of these Asian high-rise building are not of the traditional straight up and down designs but rather innovative typologies or stacks of elements. Creating models of these designs involves the careful proper placement of hundreds of elements (Figure 1).

At the Future Cities Laboratory (FCL) in Singapore and at the ETH Centre for Global Environmental Sustainability, researchers have developed model building robot systems that can construct 1:50 scale models. The model building projects help pinpoint not only issues of appearance but also issues of structural stability and construction reality.

The researchers and students at FCL have developed customized robots to accomplish the construction activity. Because building models may reach up to 4 m in height, they have adapted a Universal Robot model UR5 robot arm with 6 df on to a tall two-axis column positioning system. A family of custom-designed vacuum grippers has been developed to carefully handle the model elements (Budig et al., 2014).

Very old meets with the very innovative new in a joint project at the Bartlett School of Architecture in London and the ENSA Paris-Malaquais, Laboratoire GSA, Paris. The project focus is to automate the design and prototype a spherical vault based on a design by the work of Joseph Abeille, a French engineer, in 1699, and others. The original 1699 designs addressed the fabrication of a large span with elements much shorter than the span.

The structure is called a hemispherical Abeille’s vault. The team needed an efficient means to fabricate hundreds of wedge-shaped foam plastic pieces (Figure 2). The design software based on Rhino prepares the program for the robot to cut each piece according to the mathematical equations of the design. The programming software takes into account the parameters of the hot wire-cutting process, helping to ensure the accuracy of each piece (Schwartz and Mondardini, 2014).

Robotic building construction

Not all the robots deployed in architectural applications are just in the design studio or laboratory, many are addressing construction and fabrication challenges onsite. These challenge and address both traditional construction methods and new innovative fabrication techniques for designs. The robot has freed the architect to design structures not previously possible by traditional construction techniques.

A team at the Massachusetts Institute of Technology (MIT) has developed a truck (lorry)-mounted robot with a reach of over 24 m and a lift capability to 680 kg (Figure 3). The robot provides 11 axes of freedom. To achieve this capability, the team attached a 6-axis KUKA robot arm to a 5-axis Altec mobile hydraulic boom. The compound structure is akin to the capabilities of the human arm and hand, only very much longer.

The goal for the project is provide the ability to construct onsite non-standard architectural forms in situ, integration of real-time onsite sensing data, enhanced resolution, lower error rates and increased safety for construction personnel. Other goals include large-scale 3-D printing of insulative formwork for castable structures. The system is actively monitored with an accelerometer and a gyroscope at the end of each cantilevered element plus velocity monitoring of each hydraulic piston and angle measurements (Keating et al., 2014).

Another robot for automated construction is the result of efforts at the Bartlett School of Graduate Studies in London. As a demonstration of their efforts to deploy adaptive programming, they have constructed a robot system that can lay bricks (Figure 4). A major challenge was to develop a system that could handle unpredictable construction materials such as the mortar and work alongside human construction people.

The system combines a Universal Robots arm with Grasshopper and Rhinoceros software, feedback sensors, custom-designed end-of-arm effectors and 3-D sensors. The programmed tasks include pick and place bricks, switching end-of-arm devices, laying mortar and removing excess mortar after each new layer is in place.

Human bricklayers were monitored as part of the learning process to develop the proper mortar laying gestures. The vision system helps monitor the unpredictability of mortar and helps address getting the correct amount of mortar for each brick (Elashry and Glynn, 2014).

Complex wound fiber-based structures have become a building approach of interest. Researchers at the University of Stuttgart have developed a core-less filament winding process based on robotic technology (Prado et al., 2014). The benefit is the ability to construct a structure with high load capacity and minimal self-weight.

The fabrication system consists of a pair of facing robots working in tandem to carefully construct the filament-based structure. The 12-axes kinematic system can ensure geometric accuracy to the winding process. The project team has developed coordination software and programming techniques to accomplish the required winding patterns.

Future team goals include real-time robotic sensing to enable online modification of the robot program. This would allow the process to be adaptive and unscripted and not based on pre-determined geometry. Future developments may also address fully integrated structures with more complex geometries, weatherproofing and/or thermal insulation.

Fabrication of architectural elements

Traditionally the fabrication of architectural elements was a very labor-intensive effort. Researchers have developed many robotic applications to produce such elements more accurately and more efficiently.

At the Harvard University, researchers have developed a robotic method of creating clay and ceramic wheel thrown elements with robotic systems. Traditional ceramic elements are hand formed on a rotating wheel table. This has made it possible to create new and different designs. Throwing clay has always been a very artistic task involving the careful touch of the potter’s hands against the spinning lump of clay. The Harvard team has merged ceramic traditions, malleable materials and automation (Dickey et al., 2014).

They have applied an ABB robot to pottery tasks such as forming, carving and additive processes. By calibrating the robot arm with paper tests using a marker, they were able to gain valuable insight into the interaction of the material on the wheel and the robot arm movements.

Sound control is another factor driving architectural robotic innovation. Researchers at the University of Sydney have focused on creating complex curved structures, where the design is driven by applying complex spatial geometries to reduce sound concentration in the curved structure design. By linking these geometries directly to robot programming by using variations in a Grasshopper software and structural engineering software (Strand7) and acoustic simulation software (ODEON), they can efficiently program the robot (Reinhardt et al., 2014).

Their TriVoc system transfers the mathematical information to robot programming to produce styrofoam blocks. These blocks are then used to fabricate the full-scale structure model for testing. The robot approach enables precise control over the sculpturing of the blocks needed for the construction, helping ensure accuracy in both the full-scale test model stage and in the final structure.

Innovation in structure design and the interest in constructing thin walls has driven another robotic effort. Researchers at the EPFL Laboratory for Timber Construction IBOIS, Lausanne, Switzerland, have developed a technique to construct a thin-shelled curved structure (Robeller et al., 2014).

The prototype structure has a 13.5-m span and is but 77-mm thick and fabricated from hundreds of cross-laminated timber panels. The panels interlock on curved line joints with one another. Such panels require extremely accurate dovetail notches. Robotic cutting is the only cost-effective means to generate such dovetails or planar finger-type joints. The robot is easily programmed to accurately cut the angular and varied size dovetails.

Innovations in office design have driven research in robotic construction. Traditional offices were square or rectangular boxes with a door and maybe a window(s). More recently, the office design has moved to the open office concept, where everyone’s desk is placed in a big open hall. Workers in such office designs have often found the lack of privacy a work distracter. That has led the way to trying to develop room dividers that allow light and airflow while reducing noise and providing improved levels of privacy.

Researchers have turned to robotic room-divider building to create answers in a cost-effective manner. Such room dividers are typically constructed of thousands of small elements stacked in a semi-open design. The “wall” is frequently a single thickness of elements in an angulating curved design for strength, a very labor-intensive task if done manually.

Using digital programming tools, the robot can be directed to precisely place the thousands of elements to construct the desired room divider in a very cost-effective manner.

Creating interesting shapes and surfaces has always been an architectural aim. Now robotics provides a new and innovative means to prototype different shapes, transparencies, apertures and surface treatments. Research at the Graz University of Technology has developed a robotic thermoforming process linked with a laser cut for producing such prototypes (Weissenbock, 2014).

They have used digital design tools such as Rhino, Grasshopper and Hal with a 6-axes robot equipped with a panel holder. The prototype panel can then be subjected to hot or cold air blasts for forming and a laser cutter for creating apertures or scoring (Figure 5). The forming process heats the panel and then robotically subjects it to “deforming” against various geometric shapes and sizes of deforming stationary tools.

The robot brings precision to location, process step timing (such as heating) and very accurate repeatability. Another key advantage to the robotic forming process over traditional methods is that no custom-designed forming tool is required. The robot, as directed by the digital design programming, can direct a standardized deforming tool to create the custom deforming.

Custom-molded elements have always been an important architectural feature. Researchers at MIT have been working on ways to mass produce molds for such features using robots. Molded elements like plaster, wood and ornamental columns have been features for centuries. The MIT approach of using robots provides highly accurate mass-produced identical molds automatically (Clifford et al., 2014).

By using hot knives, the robot can carve negative geometries in expanded polystyrene (EPS) which then can be used to mold glass fiber-reinforced gypsum (GFRG) positives (Figure 6). By using two such negative molds held together, columns can also be cast. The hot knife has proved to be an excellent means to create complex surface features in the EPS foam. This approach presents an effective method for creating free from geometries without material waste and with efficiency of scale.

Architectural sheet metal cladding is another popular surface treatment approach. A joint effort by researchers at the University of Michigan and the Southern California Institute of Architecture has developed a robotic method of forming variable surface curvatures in sheet metal. They have developed methods using both single-point incremental forming (SPIF) and double-sided incremental forming (DSIF) (Kalo and Newsum, 2014).

The SPIF method uses a custom-forming end-of-arm tool on a single robot. The DSIF method uses a custom-forming tool in each of two opposing robots working in tandem. The DSIF can create panels with both positive and negative Gaussian curvatures or rapid-slope variations. The robots are also deployed to measure accuracy of the part that has been formed. By placing a laser pointer and scanner on the robot arm and moving the robot with a digital scanning routine, a data file can be gathered and subsequently used to refine the forming program.

DSIF forming presents a new set of challenges. Researchers developed two approaches to address these challenges. One method is to use a forming tool in one robot and use a support tool in the facing robot. This approach reduces inaccuracies at the edge and geometric creep. The second method uses a forming tool in each of the facing robots (Figure 7).

For information on obtaining a copy of “Robotic Fabrication in Architecture, Art and Design 2014”, by Wes McGee and Monica Ponce de Leon, the complete collection of papers presented at the 2014 Rob-Arch Conference presented at the University of Michigan at Ann Arbor, Michigan, send an e-mail to [email protected]