Mobile robot safety starts with the floor

When a mobile robot enters a facility, attention naturally focuses on the machine itself: its sensors, its software and how smoothly it avoids obstacles. Yet the safety of mobile robots is decided much earlier and much lower down — on the floor on which the vehicle and people are to move at the same time. This article shows why clear, physical boundaries on the floor are a prerequisite for every implementation, not a cosmetic extra added after the fleet goes live.

A robot does not enter a facility — it enters a designed environment

A good point of reference is the humanoid pilot BMW ran at its Spartanburg plant. The most interesting thing about this test was not that the robot walks around the facility, but that specific conclusions were drawn from earlier trials regarding safety zones, connectivity and human–robot interaction. In other words: before the machine started working, someone first had to design the space in which it would move.

This changes the way we think about implementation. A robot does not "enter" a facility on its own — it enters an environment that people have previously described with boundaries, routes and traffic rules. If that description is missing, autonomy has nothing to rely on. The machine then moves in a space that only its algorithm understands, and the people around it have to guess where its path runs and when to get out of its way. Guesswork is the opposite of safety.

Mobile robot safety starts with the floor

The floor is the first and most universal carrier of information in a facility. Everyone sees it: the operator with a forklift, the pedestrian walking between workstations, the shift supervisor and the robot itself, which reads its surroundings from a low perspective. That is why the floor is the easiest place to build a common language that people and machines understand in the same way.

A boundary drawn on the floor works independently of the fleet control system. Even if the software "knows" where the route runs, people do not have that knowledge — until they see it beneath their feet. High-contrast floor markers and floor marking tapes turn an abstract zone in the system into something that can be seen from several metres away and understood without training. This is exactly the layer that pilots such as the BMW deployment describe as safety zones — only reduced to the simplest, physical level.

A boundary you can see from several metres away

The key question for every zone is: will the operator see it early enough? If they only notice the boundary when they are already standing in it, it is not a boundary but guesswork. Effective traffic separation requires people to read the rule at a distance that allows them to react — slow down, stop, step off the path. Two layers are responsible for this legibility.

Floor marking — the information layer

Travel routes, decision points and drop-off zones are defined with floor marking. Consistency and contrast matter here: one colour for traffic lanes, another for pedestrian zones, yet another for areas of increased risk. Marking tells you where the robot's path runs and where people can move safely, but on its own it does not stop anything. It is a layer of information, not protection.

Physical barriers — the protection layer

Where marking alone is not enough — at crossings with pedestrian traffic, at workstations adjacent to the route, in narrow passages or by columns and racks that create a risk of trapping — a hard boundary is needed. Raysan flexible barriers physically separate the vehicle's path from the space used by people and protect infrastructure from the consequences of any contact. Unlike rigid guard rails, a flexible barrier absorbs the impact energy and returns to its original shape, so it protects against a single incident without needing to be replaced after every contact. The principle that orders both layers is simple: the higher the risk at a given point, the harder the boundary — from a painted line, through a sign, to a physical obstacle.

Mixed zones: where human and robot traffic cross

The most difficult places are not the straight sections of the routes, but those where the robot's traffic crosses that of operators, forklifts and manual transport. That is where risk is concentrated, and that is where human–robot interaction has to rely on simple, unambiguous signals. The operator has no time to analyse the situation — they need to know within seconds who has right of way and where the safe place is to let the vehicle pass.

Well-organised pedestrian protection addresses precisely these points: it defines crossings where routes inevitably intersect, and refuges — zones where a person can stop outside the vehicle's path. Here the boundary ceases to be merely a warning and becomes a physical separation of two streams of traffic. A well-designed mixed zone does not force the pedestrian to make a decision "by feel" — it decides in advance where to cross and where to wait.

Consistency between shifts — a boundary that does not disappear at 22:00

A safe boundary must mean the same thing on every shift and in every part of the facility. If on the first shift a yellow line marks the robot's route, and on the third someone puts a pallet on it "for a moment", the whole system falls apart. The robot will run into an obstacle nobody anticipated, and people will lose trust in markings that nobody enforces.

That is why boundaries on the floor require not only good design, but also discipline in maintaining them. Routes must be kept clear, marking must remain legible despite wear, and barriers must stay in place. This is the "boring" part of the implementation — and at the same time the part that most often determines whether the robot will be real support or a source of daily micro-collisions and stoppages.

Mini-checklist of physical boundaries before start-up

Before the fleet goes into production, it is worth checking a few simple questions that concern only the physical layer:

• Are the robot's traffic zones visible from several metres away, not only close up?
• Can an operator without training "read" the traffic rules from the marking alone?
• At the highest-risk points, has the marking been backed up by a physical barrier?
• Have pedestrian crossings and refuges been defined at traffic crossings?
• Do the boundaries mean the same thing on all shifts, and are they kept in a consistent condition?
• Does a person know what to do within seconds when their path crosses the vehicle's route?

If the answer to any of these questions is "no", then the cheapest and fastest element of the implementation — the organisation of space — is not yet ready. A robot can be delivered in a few weeks. The environment in which it will be safe is built earlier, and from the floor up.

Is the physical layer needed if the robot has its own safety sensors?

Sensors protect against contact, but they do not replace a clear boundary for people. A scanner will stop the vehicle when it detects an obstacle — but that is a reaction to a situation that has already happened. Physical boundaries and marking work earlier: they ensure that the operator does not step into the path at all, and that a risky crossing of traffic never arises. The two layers complement each other — the on-board layer protects the machine and its surroundings at the last moment, the physical layer eliminates critical situations before they occur.

Where to start when the facility was not designed for robots?

With the floor. First, fixed travel routes are defined and marked, kept away from concentrations of pedestrian traffic; next, the highest-risk points are secured with barriers; and finally, the mixed zones are organised with crossings and refuges. This order makes it possible to achieve a clear, safe environment without rebuilding the facility — before the first vehicle even appears in it.

Sources

• Automatyka.pl, "BMW tests humanoids" — conclusions from the pilot regarding safety zones, connectivity and human–robot interaction: bmw-testuje-humanoidy-249626-10