Collective robotic construction: Will machines build future cities?

BUFFALO, N.Y. — Taking cues from Mother Nature, computer science and other fields, researchers are laying the groundwork for teams of autonomous construction robots to build future cities.

The robots, which may operate similar to termites, beavers and other animals, are still likely decades away from reality. But advancements in computer coding, sensing and materials are helping push this emerging field, called collective robotic construction (CRC), into the spotlight.

The journal Science Robotics published on March 13 an overview of collective robotic construction. Its corresponding author is Nils Napp, PhD, assistant professor in the Department of Computer Science and Engineering in the University at Buffalo’s School of Engineering and Applied Sciences.

Napp and co-authors describe how the field has evolved and the challenges that lay ahead, including improving robotic autonomy, perception and design systems.

“More and more people are living in cities. Unfortunately, many of these cities lack adequate housing. Meanwhile, you have construction sites, where roughly a fifth of all worker injuries occur in the U.S.,” says Napp. “We believe collective robotic construction can help solve these problems and lead to new approaches in building and designing urban environments.”

Co-author Kirstin H. Petersen, PhD, assistant professor of electrical and computer engineering at Cornell University’s College of Engineering, says: “Automating construction could have additional applications, such as disaster relief and space exploration. For example, robots could help build temporary shelters or containment structures. They could also assist humans and other robots, like the Mars rovers, as we push further into outer space.”

The overview, which includes a companion video, explains how researchers are studying stigmergy, a biological phenomenon used to explain, among other things, how beavers build dams without blueprints. For example, the sound of gushing water indicates a leak and triggers the rodent to plug the leak with nearby materials.

Autonomous construction robots can work the same way, Napp says. This is a decentralized approach where machines operate autonomously with limited built-in instructions. As a result, such systems do not have a single point of failure and can complement approaches where a central computer gives real-time guidance to a team of construction robots.

The overview also addresses computing hardware, construction materials, performance metrics and other topics related to collective robotic construction. To make CRC a reality, the article points out areas where more research is needed:

• Autonomy — More robust algorithms are needed to guide the robots.
• Perception — Robots need better sensors to improve their understanding of the environment.
• Reliable mechanisms — Enhanced ability to move over and manipulate three-dimensional objects and environments.
• Systems integration — Better methods of combining hardware and software systems.

Additional co-authors include Robert Stuart-Smith, PhD, University of Pennsylvania and University College London; Daniela Rus, PhD, Massachusetts Institute of Technology; and Mirko Kovac, PhD, Swiss Federal Laboratories for Materials Science and Technology.