The building industry produces about 11% of global carbon emissions before anyone flips a light switch. New research from MIT suggests a fleet of small robots snapping together giant interlocking blocks could cut that number significantly, while also building faster and cheaper than the concrete methods that dominate the industry today.
Building construction is one of the last industries that looks roughly the same as it did decades ago. Workers pour concrete, bolt steel frames, and stack precast panels in sequences that are slow, labor-intensive, and carbon-heavy. Automation has reached warehouses, factories, and farms, but construction sites remain stubbornly manual.
A new study from MIT’s Center for Bits and Atoms, published April 28 in the journal Automation in Construction, offers a concrete alternative. Researchers designed a system where small robots called MILAbots (Modular Inchworm Lattice Assembler robots) crawl across a growing structure and snap together three-dimensional building blocks called voxels.
Each voxel is a lightweight geometric frame, similar in concept to a rigid, oversized LEGO brick with an open lattice structure instead of solid plastic. The blocks interlock mechanically when pressed together, meaning the robots do not need precision welding equipment, fasteners, or adhesives. They just drop a block into position and step on it.
The inchworm comparison is accurate. Each MILAbot grips the lattice with one end, extends its body, grips again with the other end, then releases the first grip and repeats. The robot uses the structure itself as its scaffold, crawling across the growing building to deposit the next block exactly where it belongs.
Why This Changes More Than It Sounds
The carbon numbers are striking. The team compared their system against three widely used construction methods: 3D concrete printing, precast concrete modules, and conventional steel framing. Steel voxels in the MIT system required only about 36% of the embodied carbon of 3D concrete printing. Wood voxels performed even better, coming in at roughly 17%.
Embodied carbon is the total greenhouse gas emitted in producing, transporting, and assembling building materials, not including the energy used to run the building once it is occupied. It is one of the hardest categories of construction emissions to reduce because it is baked into the materials themselves, not the operations of the finished building. A switch from concrete to voxel-based wood and steel assembly does not require new energy sources or policy changes. It requires a different way of building.
The speed and cost numbers hold up too. Projected on-site assembly time for the steel and wood approaches averaged 99 hours for a simple building, compared to 155 hours for existing automation methods.
The cost was competitive. The catch: one robot working alone is much slower than existing methods. The system only reaches those numbers with a distributed team of roughly 20 robots working in parallel, each handling a different section of the structure independently.
That distribution is actually a feature, not a limitation. Because the robots can work on separate sections simultaneously with no central coordination bottleneck, adding more robots scales the speed linearly. And because the voxels interlock rather than requiring bonded connections, the construction process is also reversible. A structure can be disassembled and rebuilt in a different configuration, which changes the economics of temporary structures, disaster relief housing, and modular buildings that need to adapt over time.
Where This Fits in a Bigger Shift
The voxel approach comes from aerospace. MIT’s Center for Bits and Atoms has previously used similar systems to build aircraft wing sections with Boeing and Airbus. The core insight is that a structure does not need to be monolithic to be strong. A lightweight lattice, assembled correctly, can match or exceed the structural performance of solid materials at a fraction of the weight and with far less material waste.
Applying that logic to buildings requires solving problems that do not exist in an aircraft factory, primarily that construction sites are outdoors, irregular, and uncontrolled. The MIT team has not yet tested their system at full scale in real conditions. The current demonstrations are lab-scale, and important engineering questions remain open, including lateral load stability (how the structure handles sideways forces like wind), fire resistance, and integration of plumbing and electrical routing into the lattice.
The team’s next planned testbed is a “super fab lab” in Bhutan, where they intend to use locally replicated robots to assist in construction for a planned sustainable city. That move from a controlled lab to an international field deployment is where the engineering reality testing begins.
The result to watch for: whether the system can perform comparably on an actual construction site, with all the variables a real project introduces, not just in the conditions researchers can control.
Sources
Smith, M., Richard, P., Parra Rubio, A., and Gershenfeld, N. “Comparative evaluation of robotically assembled discrete lattice systems for sustainable construction.” Automation in Construction (2026). DOI/URL: https://www.sciencedirect.com/science/article/abs/pii/S0926580526001937 Published: April 28, 2026
“Robotically assembled building blocks could make construction more efficient and sustainable.” MIT News, April 28, 2026. https://news.mit.edu/2026/robotically-assembled-building-blocks-makes-construction-more-efficient-and-sustainable-0428
Quotes in this article are drawn from the press release issued by MIT News on April 28, 2026.
More Like This
Ray Jackson holds a BSc in Electrical Engineering from the University of Manitoba and a PhD in Physics from Carleton University. His reporting interests include Current and Future Technologies, Engineering and Artificial Intelligence.