State of the Art

Raw Timber and AEC Industry

Wood waste is often disregarded in the wood industry and the construction industry. Majority of wood used in construction is sawn and processed wood that not only generates waste on its production but it is also associated with high CO2 emissions, because of the embodied energy being released during processes like: harvesting, transportation, edging and trimming. Use of saw wood is being demanded nowadays due to the overstocking of forests, affordability and minimal processing regarding edging and trimming.

Several scanning methods have been deployed and tested in order to understand the irregularities of raw wood and its design applications in the construction industry. Methods like photogrammetry, gaussian splatting, LiDAR, depth cameras and computer vision are some of the methods that are being tested to understand structural properties of raw wood, direction of fibers and irregular morphology of the wood itself. Most of the state of the art projects that are working on understanding different aspects of raw timber have been utilizing scanning methods and generating computational databases on how to design and apply for architectural elements and design for the manufacturing process. This database is an output from the scanning process and it needs to be post processed in order to be cleaned, scaled and adjusted according to its needs.


Considering the challenges of working with irregular offcuts that have been primarily or barely processed, we understand the potentially of working with this unique material allied with a robotic process that could but, process and assembly with multiple angles. The aim of this research is to develop a robotic assembly system that utilizes slab wood offcuts to produce self shading facades.


In order to achieve this objective a methodology of collecting data and material has been explored in order to design with the irregularities of this type of wood. First, a digital library is created in order to extract lenght, width and identification of boards to match with the one in the physical realm. Than this data of lenght, width and possibile angle rotation is then feedback into design to inform the scale of module, joineries and how to create self shading patterns from fully opaque to transparent. Both this informations are then applied to the design of the robotic cell and the robotic toolpath for assembly of a 1.5×1.5m prefabricated panel. Finally the panel is transported to be positioned on site.

Digital Library

To create the digital library photogrammetry was chosen in order to extract the whole tridimentionality of the boards so lenght, witdh and thickness can be then used on design phase further on. To simplify the problem of variable dimentions, the thickness of the boards were considered fixed and planarized in the planar machine to always have 2.5 cm. The boards are then scanned using Polycam and cleaned using Open3D and scaled accordinly to its real dimensions. Identification of the boards are then stored into an excel sheet as well as the lenght and width that are not only going to be utilize for design purposes but also to understand which onder to feed the feeder in the robotic cell.


From the digital library to the design the wood boards are then cutted in the middle to obtain perpendicular angles on one size and then subdvided in a range of dimensions in order to join each part using male-female joinery. A module of rotation is then designed with different angles and the populated into a facade to generate a gradient between full opaque and full transparency. In order to control the gradient, attractor points were implemented parallel to a solar radiation analysis to control self shading performance of the facade. Three rotation angles were selected to generate the self shading patterns along with different types of joinery, ranging from board to board, board to panel and panel to building.


According to the design of the facade, panels of 1.5×1.5 meters are then extracted from the facade to fit into the robotic cell and for transportation logistic. The robotic cell is both informed by the fabricated panel and its angles of position, but also informed according to the identification of each board (generated with an address coding) and lenght and width information from the digital library, Before testing the robotic cell a range of manual experiments with regular wood and wood offcuts were conducted in order to test the best angle fit of joinery. Then, in the robotic cell a dual robotic process is conducted in order to assembly the panel. A first operation to pick-cut-assembly the first module, and a second operation to then pick the module from the assembly table and nail it to the fabricated panel.

Fabrication: First Operation (Picking – Cutting – Assembling of Module)

Fabrication: Second Operation (Picking – Placing – Nailing)