A wing section already fitted to an airframe will never be placed inside an X-ray cabinet. Neither will a vessel welded into a skid, or a bonded panel several metres across. Those structures still develop internal defects, and they are still routinely released on surface indications or on a destructive cut through a sacrificial sample, because the usual way to look inside something is to carry it to a machine and shut the door.
Open-geometry robotic X-ray exists for that gap. Rather than building a larger enclosure, it removes the enclosure and sends the instrument to the object. Whether that is the right answer depends less on the technology than on the part, the site and the question being asked.
Three reasons a part stays out of the cabinet
The reasons are not interchangeable, and they lead to different decisions. A part may be genuinely too large. It may be fixed in place inside an assembly, a structure or a production line. Or it may be movable in principle but too valuable, too fragile or too heavily instrumented to disturb for the sake of an image.
Only the first is a pure size problem, and that ceiling keeps moving. Configurable laboratory systems such as the RX Solutions EasyTom line reach objects up to 720 mm long in the largest configuration, and Lumafield industrial CT was built around large parts and conveyor-fed inspection. Measure the real envelope of the parts that need scanning, not the largest thing in the building. If they fit, an enclosed system gives a cleaner result for less operational effort.
What changes when the robot carries the source
In Radalytica's RadalyX system a robot arm carries the X-ray source and a photon-counting detector around the object itself. Nothing rotates the part. The scan trajectory is a path the robot follows, so the limiting factor becomes reach and access rather than chamber volume.
The detector matters too. Photon counting registers individual X-ray photons instead of accumulating charge across an exposure on a flat panel. Inside a cabinet, source, part and detector sit in a fixed, well-characterised relationship; outside it, more of that geometry has to be recovered from the data.
Access decides most of these projects
The practical question on site is rarely whether the robot can reach the part. It is whether it can get to both sides of it. Transmission imaging needs the source on one side of the material and the detector on the other, and a welded or boxed-in structure may only offer access from one face.
Partial access is not automatically a failure. A limited angular range still supports tomosynthesis, which separates features by depth without producing the full volumetric reconstruction that all-round coverage allows, and RadalyX supports more than one imaging mode along the same robotic scan path. Establishing the real access conditions before committing to the method is what prevents a team from expecting a full reconstruction and receiving depth-resolved projections.
The logistics case is usually the stronger one
The technical argument tends to settle quickly. The one that decides budgets is operational. Sending a large structure to an external laboratory means disassembly, crating, insurance, transport, a queue, and a period with the asset out of service, all before any images exist. For an installed component, removal is often the highest-risk step of all, and it can cause damage unrelated to the defect being investigated.
Inspection that travels to the part removes that chain and makes repeat inspection realistic: the same region can be imaged before a repair and again afterwards, without the part ever leaving its position. RadalyX is used by organisations including Boeing, Blue Origin and CERN, a reasonable signal of where immovability is a genuine constraint rather than an inconvenience.
Where a cabinet system is still the right answer
Open geometry solves one problem and creates trade-offs elsewhere. For small parts at the highest achievable resolution, an enclosed micro-CT system remains the better instrument: a rigid stage, controlled magnification, stable temperature and unhurried exposure are advantages a robot working in a hangar cannot reproduce. The same holds for inspection at volume, where a fixed fixture and a repeatable recipe keep results comparable from one week to the next, and where loading a part into a chamber costs far less effort than mobilising a crew around it.
It is worth naming what no X-ray method delivers. External dimensional conformance belongs to a full-surface 3D scanner, and fast screening of large composite areas for delamination belongs to contact-free laser ultrasound. Robotic X-ray extends the reach of radiographic non-destructive testing; it does not supersede the methods working alongside it.
Radiation safety without the enclosure
A cabinet does two jobs at once, and it is easy to notice only the first. It fixes the geometry, and it contains the radiation. Take it away and the shielding has to be rebuilt as procedure: a defined controlled area, exclusion distances, warning and interlock arrangements suited to the location, trained operators, and licensing that covers work outside a fixed installation. On a shared shop floor it also means agreeing when the area can be cleared, which is a scheduling problem as much as a safety one. That is a conversation with the radiation safety officer well before the first scan. Merkel Technologies represents Radalytica in Israel and provides demonstration, installation, training and service locally.
A narrow capability worth having when the need is real
Robotic open-geometry X-ray is not a general improvement on industrial CT. It is a specific answer to immovability and should be judged that way. If everything an organisation inspects fits on a bench, a good enclosed system will serve better for longer.
The case turns when there is a recurring class of part released on partial evidence because nothing can see inside it where it stands. That has a cost, usually carried as accepted risk rather than as a line in a budget. Naming the cost tends to make the decision clear in either direction.