Brownfield automation succeeds when the operating environment is treated as part of the engineering scope.
Most industrial sites were not designed around autonomous mobile robots. They were shaped by years of production changes, new machinery, temporary storage, manual forklift routes and the practical judgement of experienced operators. That does not make them unsuitable for AMRs. It does mean that introducing robots is usually a brownfield systems-engineering project, not simply a delivery of new equipment.
The wrong question is, “Can this robot navigate our building?” A better one is, “What must change—for the robot, the infrastructure, the process and the people—to create a safe and repeatable transport service?”
Start with a real site survey
A drawing is useful, but it is not the operating environment. Walk every proposed route during normal and busy shifts. Measure clear widths at the narrowest practical points, turning space under load, slopes and transitions, floor flatness, expansion joints, drains, thresholds and surface defects. Observe temporary pallets, parked trolleys, open doors and unplanned crossing behavior.
Pay attention to time-dependent conditions. A passage may be clear at 10:00 but become a staging area at 14:00. A production aisle may be shared with a cleaning machine after a shift. The robot route must be evaluated against the behavior of the site across the working day—not only against its CAD geometry.
For every critical location, record an observation, a requirement, an owner and an action. Site readiness should be a controlled engineering register, not a collection of photographs without decisions.
Separate navigation capability from physical feasibility
A vendor may demonstrate reliable navigation in a controlled facility. That does not establish that a proposed robot can safely move a given load through a specific site. Review the loaded and unloaded vehicle envelope, protrusions, stopping distances, turning sweep, permitted gradients, required ground clearance and load stability.
Sensor visibility also matters. Glass, highly reflective surfaces, low obstacles, overhanging structures, dust, steam and changing lighting may affect different sensor systems in different ways. These should become defined test conditions rather than automatic claims that a particular sensor technology is inadequate.
For forklift-type AMRs, check pallet geometry, fork entry, floor tolerances, pick/drop repeatability and clearances in addition to route geometry. A route that fits an unloaded robot may not fit the same vehicle carrying its maximum permitted load.
Understand mixed traffic before drawing robot routes
Manual forklifts, pedestrians, trolleys and autonomous robots do not share space in the same way. Humans negotiate informally through eye contact, anticipation and experience. Robots need validated traffic rules and defined operating conditions.
Map crossings, blind corners, high-speed forklift lanes, pedestrian access, emergency exits and intersections where several processes converge. Decide where segregation is practical, where controlled shared operation is unavoidable, and where changes in signage, barriers, crossing arrangements or operating procedures are necessary. Do not treat a navigation software map as a substitute for a site-specific risk assessment.
The objective is not always to give the AMR the shortest route. A slightly longer route with fewer conflicts can deliver more predictable cycle time and a clearer safety concept.
Network coverage is an operational requirement
Ask what communications each robot function actually requires. Some AMRs can continue limited onboard navigation during temporary network loss while others rely on connectivity for mission management or fleet coordination. Those differences must be established for the chosen platform.
Survey wireless coverage under realistic operating conditions, including moving loads, metal racking, equipment states and handover between access points. Confirm firewall and IT/OT requirements before commissioning. Define the behavior during network loss: Does the vehicle stop safely, finish its current action, queue locally or require intervention? Who receives the alarm, and how is operation restored?
A successful wireless speed test in the office is not evidence of reliable connectivity along a warehouse route.
Fix pickup and drop-off logic—not just the map
Many brownfield failures originate at interfaces. Pallets are not positioned consistently. Operators use a temporary buffer that is not reflected in the WMS. Pickup stands change location. Empty return flows are undocumented. A robot cannot reliably compensate for a process that lacks a clear definition of what “ready to pick” means.
Standardize transfer locations and tolerances. Specify who makes a load ready, when the system creates the mission, how successful transfer is confirmed, and what happens when the expected pallet is missing or badly positioned. Review doors, lifts, conveyors and interlocks as active system interfaces rather than incidental building features.
Prioritize changes by impact
Not every imperfection requires a construction project. Classify findings into four decisions:
Accept and validate: an existing condition fits the documented operating envelope and passes a representative test.
Modify the operation: adjust staging, traffic rules, housekeeping or responsibilities without major civil works.
Modify the infrastructure: repair the floor, improve a doorway, add a handover fixture or upgrade the network where the business case justifies it.
Reject or redesign the concept: if a hazard, physical limit or process mismatch cannot be adequately resolved, change route, robot class, automation scope or project timing.
The fourth option is a legitimate engineering outcome. A responsible feasibility study must be able to say that an application is not ready.
Commission in stages
A practical rollout can begin with one representative route and a limited operating window. Validate safe behavior, communications, handling accuracy, traffic interactions, alarms and recovery. Extend the operating hours and mission complexity only after the evidence supports it.
Before go-live, ensure that operators know what they may and may not do around a robot, maintenance teams have defined access and isolation procedures, and shift leaders understand ownership of exceptions. A commissioning plan should include contingency procedures for critical material flows.
Brownfield readiness checklist
- Routes surveyed under representative shifts and traffic conditions
- Floor, slopes, clearances and loaded turning envelopes verified
- Pick/drop locations and handling tolerances standardized
- Site-specific safety concept and mixed-traffic rules documented
- Communications surveyed and failure behavior agreed
- Doors, lifts, WMS and handover interfaces assigned owners
- Infrastructure modifications costed and prioritized
- Phased acceptance and operational fallback defined
The engineering conclusion
A factory does not have to be newly built to use mobile robotics. But it does need a clear understanding of its existing constraints and the willingness to make targeted changes. A good brownfield project adapts the technology and the operation to each other—rather than expecting one side to absorb every weakness of the other.
ČAMINSKI — Engineering Advisory & Project Delivery. For brownfield AMR feasibility and site-readiness reviews: office@dimitarcaminski.com.