Building Momentum

The age old profession of construction is increasingly leveraging modern technology to complete projects faster and more efficiently, and to reduce costs. One of the more interesting innovations is the use of gantry robots — these massive systems perform pick and place operations, welding, or even 3D printing of concrete throughout the entire structure being built. While these robots can greatly speed up projects, they must be very large, and their size scales superlinearly as the size of the structure being built increases. For larger structures, that quickly becomes unmanageable. The cost of these systems is also very high, which limits when and where they can be put to work.

To be more generally applicable and really transform the industry, a more adaptable approach is needed — something that can handle large jobs without itself growing to mammoth, unwieldy proportions. Researchers at the Massachusetts Institute of Technology, in conjunction with the U.S. Army Research Laboratory, have taken a cue from microbiology to develop a novel robotic construction system. In a way similar to how ribosomes turn aminos acids into complex protein structures, which in turn make up organelles, cells, and ultimately entire organisms, their technique uses small building blocks that make up much larger structures. Another very interesting feature of their technology — again much like ribosomes — is that their building blocks can recursively make more building blocks.

The basic unit of the self-replicating robot assembly system is a voxel — a three-dimensional pixel — which is roughly cube shaped and made with a lattice structure. These voxels are not only mechanical structures, but the connections between them can also carry both power and data to allow for coordination between units. With a few of these voxels joined together, the larger structure performs tasks like lifting, moving and manipulation of materials. And once a voxel-based robot has a minimum set of capabilities, it can also build more robots. Those new robots could be copies of itself, larger robots, or robots with new abilities.

As a swarm of robots went about building a structure, there would come a time when they were traveling larger and larger distances to place the next voxel on the growing structure. When that begins to cause efficiency problems, the robots could pause their construction work, and instead temporarily focus on building a larger robot that is better suited to the job as conditions change. Likewise, if fine, detailed work is needed after the initial rough work of building, for example, walls, the large robots could build smaller machines that can better handle the job.

To manually program the robots to operate in this way for each job would be challenging and labor intensive, so the team designed an algorithm that determines when to build additional helpers, and what sort of robots those helpers should be. This modular, autonomous, self-replicating approach completely sidesteps lengthy retooling periods that are typically needed before building a new structure.

The researchers acknowledge that their work is still in the early stages, and it will be some time yet before methods such as these can be used in the real world. At present, they are fine-tuning some aspects, like the force of the actuators and the strength of joints, with the goal of making their technology a practical option for industrial use.