
The UK Atomic Energy Authority’s robotics team and the operator of the Large Hadron Collider (LHC) have developed a novel approach to maintaining the 27km-long facility: robotic mice.
The 3.7cm-wide robot is designed to travel autonomously through long, narrow pipes called beamlines, along the length of the LHC.
At the heart of the LHC, the beamlines are surrounded by superconducting magnets kept at -271°C. The beamlines operate under ultra-high-vacuum conditions, and their position deep within the infrastructure that supports these extremes makes human access and inspection extremely challenging.
To meet these challenges, CERN (the LHC operator) partnered with the UK Atomic Energy Authority’s robotics centre RACE (Remote Applications in Challenging Environments) – which has remote handling expertise in hazardous and hard-to-reach environments – to develop a robotic solution.
Together, the teams created the 20cm-long robotic mouse, named PipeINEER. Unlike existing pipe inspection systems, PipeINEER can navigate up to 6km on battery power alone on a single mission while operating in a space only a few centimetres wide.
AI trained on LHC imagery
As the robot moves, it captures detailed images of each of the 2,000 plug‑in modules (that handle the expansion and contraction caused by the harsh conditions) and uses AI trained on real LHC imagery to detect any abnormalities. It is equipped with energy‑efficient systems and multiple safety features that monitor its performance during long autonomous runs.
If the robot detects an issue, it returns to its starting point and reports the location of the problem. This approach allows engineers to address specific points along the LHC, without disassembling large sections of pipe and using a manual endoscope – a process that is time-consuming and expensive.
The robot will be tested over 60km of operations later this year. Following this testing, final units will be manufactured towards the end of the year, and CERN operators will be trained on the new units in early 2027.














