Putting Pompeii’s pieces together with a robot’s help

Pompeii’s ancient Roman frescoes, disintegrated and buried for centuries, could be given a second life thanks to a pioneering robotic system designed to support archaeologists in one of their most challenging tasks: reassembling shattered artefacts.

Developed under an EU-funded project called RePAIR, the technology combines advanced image recognition, AI-powered puzzle solving and ultra-precise robotic hands to speed up traditionally slow and often frustrating restoration work.

The robot project exhibited in Pompeii, launched in 2021 and coordinated by Ca’ Foscari University of Venice, brought together international research teams that used the archaeological site as a testing ground.

Gabriel Zuchtriegel, the site’s director, said the experimental project “actually started from a very concrete necessity to recreate fragments of frescoes destroyed during the Second World War.”

Researchers believe the technology could transform restoration practices worldwide.

The robot uses twin arms equipped with flexible hands in two sizes and vision sensors to identify, grasp and assemble parts without damaging their delicate surfaces.

The once-thriving city of Pompeii, near Naples, and the surrounding countryside were buried under volcanic ash when Mount Vesuvius erupted in AD 79.

Researchers focused on frescoes preserved in fragmentary form in Pompeii’s repositories; Damaged during the first explosion and later destroyed during World War II. Two large ceiling paintings were destroyed by bombing in World War II, and the so-called House of Gladiators frescoes collapsed in 2010.

Replicas were created during this initial testing phase to avoid compromising the original parts.

While robotics teams worked on the design and construction of the system, artificial intelligence and machine learning experts developed algorithms to reconstruct the frescoes and match colors and patterns invisible to the human eye.

Experts say the task is akin to solving a giant puzzle with extra challenges such as missing pieces and no reference image of the final result.

“It’s like you buy four or five boxes of puzzles. You mix everything up, then you throw away the boxes and you try to solve four or five puzzles at the same time,” said Marcello Pelillo, a University of Venice professor who coordinated the project.