Goldberg Robotics

[Image: From Science Daily/University of Oslo].

Robots emitting robots emitting robots: this is one way that machines will learn to navigate extreme spatial environments.

"In the future," we read in a press release courtesy of Science Daily, "robots must be able to solve tasks in deep mines on distant planets, in radioactive disaster areas, in hazardous landslip areas and on the sea bed beneath the Antarctic"—as well as in the cracks of otherwise inaccessible archaeological sites.

Researchers at the University of Oslo think we need to send machines capable of not exactly of replication, but something more like budding or fruiting, using 3D printers.

Kyrre Glette, one of the researchers behind the press release, imagines a robot being sent into "the wreckage of a nuclear power plant," for example, where it encounters a stairway it had not been anticipating needing to climb. For the moment, it's stuck. So what does it do? "The robot takes a picture. The picture is analysed. The arms of one of the robots is fitted with a printer. This produces a new robot, or a new part for the existing robot, which enables it to negotiate the stairs."

The original robot—which was thus not single but a crowd waiting to happen—moves forward through the landscape by sending detached variations of itself further ahead. You could think of it as Goldberg robotics: advancing through variation.

This is obviously not a new vision—the idea of 3D printers printing 3D printers that can 3D-print further 3D-printer-printing 3D printers, for example, is a long-running staple of stoner sci-fi. Nonetheless, it's interesting to see this specifically discussed in terms of navigating spatial environments, be they mines, caves, or architecture, explored and mapped by an instant machine-ancestry self-produced specifically for the task at hand.