dormakaba, ANYbotics and LEGIC have demonstrated a way for autonomous inspection robots to pass through access-controlled doors using digital credentials, removing a practical obstacle that can limit unattended inspection routes inside industrial plants. The pilot was carried out at GE Vernova’s Whitegate power plant in Ireland and is now being developed towards a commercial add-on for the ANYmal robot fleet.
The test connected an autonomous robot to the site’s existing access-management environment rather than creating a separate route or leaving controlled doors permanently open. ANYmal was issued with a trusted credential that could be authenticated by the access-control infrastructure, allowing it to approach a protected doorway, present its identity, receive authorisation and continue its inspection mission without an operator opening the door manually.
That apparently modest change addresses a significant weakness in autonomous inspection. Industrial sites are deliberately divided into controlled zones, with doors and gates used to restrict movement according to safety, security and operational requirements. A robot may be capable of negotiating stairs, uneven flooring and dense process equipment, but its autonomy is curtailed if every secure doorway requires human intervention.
ANYbotics’ ANYmal platform is designed for autonomous inspection in complex industrial environments. The robot combines visual and thermal cameras, an ultrasonic microphone, LiDAR and integrated lighting with four-legged mobility, while additional payloads can be fitted for tasks such as gas detection and acoustic imaging.
Those sensors allow the platform to collect equipment-condition data during repetitive inspection rounds, reducing the need for technicians to visit every measurement point manually. Access control, however, remains part of the building rather than the robot. Unless the two systems communicate, a route that is technically navigable can still be operationally inaccessible.
The Whitegate pilot addresses that boundary by treating the robot as an identifiable user of the site rather than an exception to normal security procedures. Its credential can be governed within the access-management system, allowing permissions to be assigned according to the areas the machine is authorised to enter.
That model gives plant operators more control than simply allowing an autonomous system unrestricted movement. Access could be limited to defined zones, operating periods or inspection tasks, while events are logged through the existing security infrastructure. Robot movement can therefore remain auditable rather than relying on doors being mechanically bypassed for the sake of automation.
The integration also links two areas of industrial technology that are usually engineered separately. Autonomous inspection combines mobility, mapping, sensor processing and fleet management, while physical access systems govern doors, credentials and permissions. Bringing the two together means the robot’s identity becomes another part of the operational technology environment.
That creates additional engineering questions. Robot credentials have to be issued, renewed and revoked correctly; permissions need to follow changes in inspection routes; and the system needs defined behaviour if communications fail or a doorway cannot be opened. Access to hazardous or restricted areas also remains subject to plant-safety requirements irrespective of whether the robot possesses a valid digital identity.
The commercial value will therefore depend on how comprehensively those controls can be managed. Authentication at one door is straightforward compared with deploying autonomous machines across a large industrial site containing multiple security zones, lifts, gates and areas governed by different operating procedures.
The partners are already looking beyond conventional doors. Their next development work includes lifts, gates and turnstiles, which could extend autonomous routes through larger and more complicated facilities. Lifts are particularly relevant in multi-storey plants, where a robot otherwise needs dedicated infrastructure or manual assistance to move between floors.
Improved access does not remove the need for people from inspection and maintenance. Robots can collect thermal, visual, acoustic and other condition data, but technicians still have to interpret many findings, plan interventions and carry out repairs. Autonomous inspection is most useful where it removes repetitive data-collection work and increases the frequency with which equipment can be observed.
Physical access is one of the less visible constraints on achieving that objective. A machine capable of walking for hours without supervision gains little if its route repeatedly stops at locked doors. Giving it a governed identity allows existing security infrastructure to become part of the automation architecture rather than a barrier to it.
The Whitegate deployment is therefore significant less for the individual door opening than for the operating model behind it. The robot remained subject to plant access rules rather than being given a special mechanical workaround, preserving the principle that autonomous equipment should operate inside the same controlled environment as other authorised users.
If the integration can be extended reliably across doors, lifts and gates, digital identity could become a standard component of autonomous inspection projects. Robots would still require mapping, sensing and task configuration, but secure movement between plant areas would no longer have to depend on an employee following behind with an access card.
