When Connected Machines Must Know They Cannot Continue Safely
An autonomous vehicle can lose confidence in what it sees. A warehouse device can depend on a wireless connection that supports its work. In both environments, the security and safety question extends beyond whether the technology functions under normal conditions: what happens when a critical input becomes unreliable?
Written responses supplied to The Supply Chainer by Seyond, Greenet and AIT Worldwide Logistics expose different parts of that challenge. Their observations concern sensing, warehouse connectivity and commercial trucking. Together, they show why operators need to understand the limits of connected automation before making it essential to daily operations.
Detection Must Include Its Own Limits
Oliver “Oli” Ramoli, Business Development & Manager Key Accounts EMEIA at Seyond, a LiDAR technology provider, said autonomous trucks should be tested as complete perception systems in their actual working environments. Those tests should include adverse weather, changing light, blind spots and interactions with people and vehicles.
Stationary, low-profile and partially hidden obstacles require particular attention. Detecting a hazard is insufficient if the vehicle cannot identify it early enough to stop safely.
“But there's another aspect that is just as important: the vehicle needs to recognize when its perception is no longer reliable enough to continue operating safely,” Ramoli said.
Dust, mud and water droplets can impair optical sensors. His recommendation is to adapt inspection and cleaning to site conditions, supported by diagnostics where available. The response did not specify detection thresholds or the vehicle behavior triggered by degraded sensing. Those remain important questions for an operator evaluating a particular deployment.
Connectivity Becomes Part of the Operating Dependency
Greenet’s warehouse example illustrates how communications infrastructure supports moving equipment. In its written reply, Jurjen Veldhuizen, the company’s CTO, described a deployment at Logent in Bleiswijk, the Netherlands, using dedicated spectrum and SIM cards.
“The network supports handheld scanners for the order pickers and AGV type of operations, enabling a stable connectivity inside the whole warehouse without hand-over issues.”
Veldhuizen said communications were encrypted and Greenet managed the network. His response did not specify the wireless generation used at the site, provide measured availability or explain how connected equipment responds to a network interruption.
Encryption addresses a different concern from continuity. It does not, by itself, establish whether a device can complete a task, retain a safe operating state or recover correctly when communication is interrupted. The Logent account demonstrates a stated operational dependency, rather than evidence of a tested response to failure or attack.
Warnings Must Also Work for People
AIT Worldwide Logistics raised a further issue: automated systems operate around people who may not understand their signals. Mike Rothacher, Executive Vice President, Road, questioned how motorists would interpret warning beacons associated with autonomous trucks.
“In the absence of a broad educational campaign with elements like public service announcements, the general driving populace is unlikely to understand beacon alerts and perceive them as potential roadside hazards.” Rothacher also identified schedule changes and navigation in densely populated areas as challenges for direct delivery. His comments concern public-road operations, distinct from Greenet’s warehouse deployment, but reinforce the importance of the surrounding operating environment.
These responses do not establish a common cyber vulnerability or demonstrate that any named system is unsafe. They identify boundaries that security and operations teams need to examine together: unreliable perception, interrupted connectivity and warnings that people may misunderstand.
The practical readiness test is whether those boundaries produce a defined response. Operators need to know when equipment should stop, who receives the exception and what must be checked before work resumes. Normal operation is only part of the assessment.
