Defining the AGV Selection Space #
An automated guided vehicle is not simply a self-driving cart. It is a coordinated system of navigation sensors, safety scanners, drive and steering hardware, energy storage, and a fleet-level traffic controller that must coexist with operators, racking, loading docks, and an existing material flow architecture. Selecting the wrong vehicle technology is often less expensive than selecting the wrong system boundary, but both are costly. This article separates the selection criteria that matter at the proposal stage from the application boundaries that will govern daily operation, and explains how each component contributes to the symptoms seen on the warehouse floor.
The decision to introduce an AGV fleet should begin with a truthful statement of the task. The task includes the number of pallet movements per hour, the distance travelled, the route mix, the handling interface at pickup and drop points, and the dwell time at each location. Once the task is fixed, the vehicle type, navigation method, charging strategy, and safety configuration can be evaluated. When the task is only loosely defined, or when route changes are expected at high frequency, the application boundary has already been exceeded regardless of the vehicle’s advertised flexibility.
Load, Floor, and Route Boundaries #
The mechanical selection criteria for an AGV are usually stated as payload, lift height, and drive type. In practice, the load handling interface and the floor condition determine whether the vehicle can operate at its design rate. An AGV that is rated for 1,500 kg but must pick up loads from a roller conveyor with a misaligned stop will not reach its rated throughput. The interaction between the vehicle load handler and the upstream or downstream conveyor, lift table, or rack position is the first boundary that must be measured, not assumed.
Load Handling Interfaces #
Fork-based AGVs require clearly defined pallet presentation tolerances. The vehicle’s fork cameras or position sensors must obtain a consistent frame of reference for each pickup. If the pallet is offset by more than the vehicle can compensate for, the controller will either stop or attempt a chase that may damage the pallet or the rack. The selection criterion for the load-handling option should therefore be based on the measured pallet placement spread at every pickup and drop zone, including under loaded conditions when the floor may deflected.
Conveyor-based and tugger-style AGVs shorten the decision boundary differently. A tugger train requires trailers to be coupled manually or automatically, and the coupling interface becomes a daily maintenance point. A conveyor deck requires exact height matching with the station, and any differences in conveyor level become observable as late pickups, dropped loads, or sensor false triggers. All of these symptoms are often recorded as vehicle faults, when the true cause is the boundary between the vehicle and the handling furniture.
Floor and Route Geometry #
AGV navigation relies on a predictable relationship between the vehicle drive wheels and the visual, magnetic, or reflector landmarks used for positioning. A floor that changes height by 10 mm across a joint, a surface with standing water, or a racking area with frequent dust accumulation will slowly degrade that relationship. The observable symptoms are often intermittent: the vehicle drifts a few centimetres at the same location once per shift, then recovers; a safety zone triggers at a gap that has existed for months; or the vehicle approaches a pickup station with a noticeable angle correction that slows the process.
Route geometry also includes the width of aisles relative to the vehicle’s steering capability. A three-wheel fork truck with a large turning radius may navigate a 3.2 m aisle in theory, but if the fleet traffic controller must recover clearance deviations for every turn, the actual throughput will drop. The selection boundary should be defined by the corridor width after the vehicle’s forward and lateral clearance zones are subtracted, not by the nominal aisle width. The clearance zones are functions of the navigation tolerance, the safety scanner field width, and the type of goods carried.
Navigation Technology and Its Boundary Conditions #
Every navigation technology has a working principle that creates a specific failure envelope. The boundary is not which technology is more “modern”, but how each method responds to changes in the operating environment. The table below summarises the practical boundary conditions, observable symptoms, and maintenance evidence for common AGV navigation types.
| Navigation type | Boundary condition | Observable behaviour when violated | Evidence to collect |
|---|---|---|---|
| Magnetic tape | Tape continuity and magnetic field strength; no metal cut-outs near the tape path | Vehicle overshoots at the same point; corrective steering oscillation; route loss after the tape patch | Magnetic field meter readings along the track; comparison of tape condition at failure point vs healthy points |
| Laser reflector | Unobstructed line of sight to at least three reflectors; stable reflector positions | Position lock loss at the same area; long initial localisation delay; intermittent “no reflector” errors | Reflector map logs; cleanliness of reflectors; presence of new storage or hanging loads near the beam path |
| Natural feature / SLAM | Stable and sufficient features in the environment; lighting consistency; no large blank walls | Progressive drift over a route; different path shape during night shifts vs day shifts; localisation divergence after racking changes | Localisation scores from the fleet controller; comparison of map feature matching; timeframe correlation with lighting or rack movements |
| Inductive wire | Continuous wire integrity; no breaks introduced by floor repairs | Vehicle stops or loses steering at the exact break point; behaviour repeats every cycle | Wire continuity testing; floor repair log; correlation of fault address with physical location |
The selection criterion for navigation should therefore be derived from the environment’s rate of change. A warehouse that regularly changes rack configurations, supports seasonal pallet flow patterns, and reallocates storage zones will push natural feature navigation into a boundary where the map must be updated frequently. A facility with stable aisles and predictable routes may be better served by a lower-cost magnetic or inductive solution. The cheapest navigation system is the one that matches the expected rate of environment change, not the one with the lowest acquisition price.
Fleet Traffic and System-Level Interaction #
Fleet-level interactions are the most commonly under-specified part of an AGV selection. The vehicle’s local controller is responsible for its own path and safe stopping, but the fleet controller assigns destinations, coordinates intersection priority, manages traffic zones, and arbitrates charging. When a warehouse problem is described as a “slow AGV”, the cause is frequently the traffic management logic rather than the vehicle itself.
Traffic zones can be designed as areas, wait points, or path reservations. With high vehicle density, the selection boundary becomes the ratio between the number of active vehicles and the number of physical routes. A system that works well with three vehicles and two intersecting corridors can quickly degrade to zero throughput with six vehicles because every vehicle is waiting on a traffic node. The observable symptom is not a fault code, but a pattern of prolonged idle time at the same intersections. This can be confirmed by reviewing the fleet controller’s task history and vehicle state logs over a full shift, measuring dwell time at known wait locations.
Component interactions also occur at the equipment handshake level. An AGV that interfaces with a high-speed door must receive a door-open confirmation signal before crossing a threshold. If that signal is unstable, the vehicle will stop in front of the door with no obvious obstacle. Similar situations occur at elevator control interfaces, barrier interfaces, and conveyor integration points. These are not vehicle failures but interface boundary violations. The evidence collection process should include a log of the external controller state at the moment the AGV stops, not only the AGV’s own fault data.
Safety Interfaces and Perception Zones #
Safety is the boundary that cannot be optimised away. AGVs use safety-rated laser scanners, bumpers, emergency stop buttons, and external field switches to protect personnel and infrastructure. The selection criteria here are the response time, the protective field size, the capability to reduce the field at corners, and the mechanical integrity of the bumper mounts. Application boundaries, however, are determined by the installed environment, not by the safety specification.
Observable symptoms of safety system boundaries are usually consistent and repeatable. They may be false emergency stops, frequent slowdowns in a particular aisle, or a vehicle that always brakes when passing a specific racking corner. The fastest way to misinterpret a safety fault is to assume that something is physically present in the vehicle’s path. The evidence collection routine should distinguish between a sensor hardware issue, a contamination issue on the scanner window, and a genuinely occupied protective field.
Diagnostic Evidence For Safety Scanner Events #
- Record the timestamp of the emergency stop and the vehicle position from the fleet controller log.
- Retrieve the safety scanner’s recorded field occupancy data, if available from the OEM interface, and compare it with camera or operator observations at the same moment.
- Inspect the scanner window for dust, grease, or condensation. Contamination frequently causes intermittent nuisance stops that occur only when the scanner rotates or when the vehicle passes through a specific draft.
- Check for new material, pallets, or storage that may have intruded into the nominal protective field without being reflected in the path planning map.
- Compare the event log with the previous maintenance actions, including scanner alignment adjustments, reflector replacement, and localisation map changes.
Safety-related events must always be treated with a conservative posture. Site procedures, lockout requirements, OEM documentation, and competent engineering judgment take priority over any attempt to bypass or attenuate protective functions. The purpose of the diagnostic routine is to identify whether the safety boundary has been physically exceeded, or whether the system’s perception of the boundary is disturbed by environmental factors. Both cases require action, but the action is different: the first is an operational change, the second is a maintenance or installation correction.
Charging Strategy and Energy Boundaries #
The energy infrastructure of an AGV operation is frequently an afterthought in the selection process, yet it is often the source of the most confusing intermittent faults. Charging may be executed automatically through contact pins, inductive pads, or opportunity charging at workstations. Each method has its own boundary conditions: pin alignment, pad cleanliness, connector wear, and battery state traceability.
A common observable symptom is the “charge hunt”, in which the vehicle repeatedly approaches a charging station, begins a charge cycle, then reports a charge fault and moves to another station. This can be caused by a damaged connector, an grounding issue that trips a residual current device, or a fleet management rule that does not allow the battery to discharge deeply enough to accept an opportunity charge. A related symptom is the vehicle that performs normally at the start of a shift but begins to slow down and lose positioning accuracy later in the day, suggesting voltage sag under load rather than a navigation failure.
Evidence collection for energy-related issues should include the battery state at the start and end of each shift, the charging current and duration at each charge event, and any correlation between charging faults and other vehicle errors. A simple indication is the number of recharging events per vehicle per day. If a vehicle charges twice as often as its fleet peers, the issue is likely not the charger itself but the energy draw, which can be caused by software settings that keep the safety system in a high-current mode, by drive motor drag from a brake that is not fully releasing, or by a navigation routine that makes the vehicle perform excessive correction manoeuvres.
Common Interpretation Errors in AGV Diagnosis #
Maintenance teams familiar with fixed machinery often apply a single-fault mindset to AGV systems, which are distributed systems by nature. Several patterns of incorrect interpretation are common in warehouse operations.
Misreading Localisation Faults as Drive Faults #
An AGV that weaves down the aisle is often reported as a steering problem. The drive and steering assembly may be perfectly healthy. In many recorded cases, the weaving is the vehicle steering to match a faulty localisation estimate, especially when a reflector is displaced or a floor feature is poorly mapped. The correction belongs in the navigation calibration, not the drive mechanics.
Blaming the Vehicle for a Traffic Coordinator Decision #
A vehicle standing still for three minutes at an intersection may appear stuck. The vehicle is not stuck; it is waiting for permission from the traffic controller. The evidence is in the fleet controller software, not in the vehicle’s brake status. Without reviewing that log, the maintenance engineer could replace the vehicle controller while the actual issue is an unset zone in the traffic logic.
Confusing Contamination with Obstacle Presence #
Safety scanner false triggers are commonly caused by a thin film of dust on the scanner window, especially in dry warehouse environments. Replacing the scanner is unnecessary and expensive. The routine check should first clean the window and run a controlled operational test before considering hardware replacement.
Attributing Battery Degradation to the Charger #
When a battery reaches the end of its service life, the vehicle may show intermittent power issues that seem unrelated to the battery. For example, the navigation system receives voltage below its operating threshold during and immediately after a pallet lift, which causes a position loss. The vehicle reports an ambiguous error, and the engineer may initially suspect the onboard power supply. A review of the cell voltage profile during the same manoeuvre is necessary to separate the battery boundary from the power conversion boundary.
Maintenance Implications and Evidence Discipline #
AGV systems do not fail according to the calendar as predictably as conventional conveyor systems. The boundary between selection criteria and application constraints must therefore be tracked with structured evidence. Time-based maintenance is still valid for tyres, castors, scanners, and connectors, but condition-based maintenance is often more effective for localisation components and charging interfaces.
A practical evidence collection method is the weekly fault log review. The fleet controller can record a list of fault events for each vehicle, including the timestamp, the fault code, and the location. The maintenance engineer should filter this by message class and by location, then look for correlations. A fault that always occurs at the same map position is almost certainly an environmental or infrastructure issue. A fault that occurs at random positions but only during high-activity periods is likely related to traffic management or battery state. A fault that occurs across multiple vehicles at the same physical location points to a floor problem or a structural change in the building, such as a new air-conditioning duct producing interference or a newly installed shrinkwrapper that blocks a reflector.
Maintenance records should also include the floor condition. Many AGV route issues are introduced by floor repairs, expansion joint modifications, or pallet drop damage that occurs overnight. A dated photographic record of the route, especially at turns and pickup points, helps distinguish between a progressive degradation and an abrupt event. The same photographs can be used when deciding whether a floor resurfacing or an underlayment patch is required before the vehicle can operate reliably.
Application Boundaries: When the AGV Is Not the Answer #
The final selection criterion is honestly identifying the application boundary at which an AGV becomes a poor technical solution. The most common boundary is the requirement for high path flexibility with extremely tight response times to spontaneous routing changes. A manual forklift can accept a new destination in seconds; an AGV must be re-tasked through the controller, the navigation path must be validated, and traffic rules must permit the new route. If the operation changes pick paths more than several times per shift and the change must take effect immediately, the AGV’s strength in repeatability becomes a liability.
A second boundary is the presence of uncontrolled human traffic. Although safety scanners and controllers handle interactions, the throughput of an AGV degrades when it must constantly stop and restart to maintain a safe separation distance. In mixed-traffic areas with narrow aisles, the selection boundary is not whether the vehicle can stop safely, but whether the number of human-pedestrian interactions will reduce the duty cycle below the required throughput. A direct observation of the current manual or semi-automated traffic density should be part of the decision process.
A third boundary is the absence of a stable route and station layout. If the material flow is still being reorganised, the AGV system will require repeated reconfiguration of maps, traffic zones, and station coordinates. While natural-feature navigation has reduced the effort of remapping, the fleet controller logic still requires careful validation after every layout change. The cost of layout flexibility must be included in the total cost of ownership, not hidden in the maintenance budget.
Finally, the decision boundary must consider what happens when the AGV fails. In a conventional warehouse, a broken conveyor can be bypassed by manual handling. With an AGV system, a single broken vehicle blocking a main aisle can stop all other traffic. The selection process must therefore include a recovery plan: where the vehicle can be manually towed, how the fleet controller handles a halted vehicle, and who decides whether the remaining vehicles continue operating around it. The absence of a recovery procedure is arguably a more critical application boundary than the vehicle’s throughput capability.
Key Takeaways #
- Selection starts with the measured task profile and the measured environment, not with the vehicle brochure. Positioning tolerance, floor condition, aisle width, and the rate of layout change define which technology can survive daily operation.
- Navigation faults are often environmental. If a fault repeats at the same map location, inspect the route, the floor, the reflectors, the tape, and the lighting before replacing any vehicle hardware.
- The fleet controller is part of the system. Standstill events and slow-downs are frequently the result of traffic management decisions, external controller states, or handshake signals, not vehicle failures.
- Safety scanner events must be treated carefully. Collect the scanner’s recorded data, inspect the window, and control the environment before considering a sensor replacement. Never bypass or disable a safety device; follow site procedures, lockout requirements, OEM documentation, and competent
Related Pearl Gateway Guides #
