The Robot That Builds Where Nothing Else Can

Most construction robots need a factory. MIT's newest prototype needs a flat surface.

Researchers at MIT's Center for Bits and Atoms have built a robotic assembly system using modular lattice blocks called voxels — think LEGO, but structural-sized and assembled by robots that crawl across the growing structure. The pitch: build anywhere, with nothing but the blocks themselves and whatever flat ground exists. No crane. No factory floor. No established supply chain. The next test of that claim is happening in Bhutan.

The system pairs voxels with robots called MILAbots (Modular Inchworm Lattice Assemblers) — inchworm-style machines that grip voxels at each end and crawl across the structure as it rises. The interlocking geometry handles the structural load. No bolts, no adhesives.

The carbon numbers are real and they are large. Compared against 3D concrete printing, precast concrete, and steel framing, the voxel system could cut embodied carbon by up to 82 percent and trim average assembly time from 155 hours to 99, MIT News reported. The catch: those numbers depend heavily on material choice, and the system has never been built outside a lab. The question is whether "build anywhere" translates into a genuine deployment capability, or whether it just means the lab benchmarks look better than they would on a real site.

The answer, if it comes, will come from Bhutan. The next testbed is planned for the Gelephu Mindfulness City project, a $100 billion planned sustainable city in the Himalayan kingdom where no crane or concrete supply chain yet exists, Construction Briefing reported. Year two of a two-year partnership agreement is running.

"We are taking aerospace principles and applying them to buildings," said Neil Gershenfeld, MIT professor and director of the Center for Bits and Atoms, in an MIT News interview. "Why don't we make buildings as efficiently as we make airplanes?" Gershenfeld's lab has spent years applying voxel assembly to space structures, wind turbine blades, and aircraft components with NASA, Airbus, and Boeing. Construction is the next application.

Voxels are modular 3D subunits that interlock into larger structures. The MIT team evaluated eight existing voxel designs, then developed a new geometry based on an octet lattice: a high-strength, self-aligning structure that requires no connectors. The pieces interlock mechanically, according to the arXiv preprint.

"The interlocking nature of these voxels means we can get nice mechanical properties without needing a lot of connectors in the system," said Miana Smith, lead author and CBA graduate student, in the MIT News interview. "The construction process can run a lot faster."

The team tested voxels made from three materials: plastic, plywood, and steel. The carbon numbers vary significantly. Plywood voxels required roughly 17 percent of the embodied carbon of 3D concrete printing and 24 percent of precast concrete. Steel voxels came in at 36 percent and 52 percent respectively. The headline 82 percent figure applies only to the best material choices against specific conventional methods.

One MILAbot working alone is slower than conventional techniques. The team's calculation shows 20 robots working in parallel is the threshold where the system catches up to or surpasses existing automation at lower cost. No one has built a structure with 20 MILAbots yet.

Scalability, durability, and fire resistance remain unexplored, the paper acknowledges. Gershenfeld frames the work more carefully: "Our work helps support why doing this type of distributed robot assembly might be a practical way to bring digital fabrication into building construction." Might be.

Reversibility is a genuine advantage no conventional construction method offers. It requires the voxels to hold together under load, through weather, and without the fire resistance that most building codes currently require.

Bhutan is the load-bearing test of that claim. The Gelephu site is semi-abandoned agricultural territory on the jungle edge, no established supply chains for cranes or concrete. An airport groundbreaking and volunteer community work are underway. It is not a pristine blank slate, but it is also not a place where conventional construction infrastructure is already in place. Whether voxels can build where nothing else can, and hold together while doing it, is the question the next two years will begin to answer.