AGV pilot: how to launch your first robots without stopping production
The installation phase is the moment when a project on paper turns into vehicles driving around the shop floor. For a plant with no automation experience, the safest route is to start small — an AGV pilot in one area, with a few vehicles, before the system covers the whole of production. This guide walks you through six practical elements of this stage: from installing the equipment, through choosing the pilot scope and measuring results, to the criteria for moving into continuous operation. It is the sixth step of the step-by-step AGV implementation methodology — it follows preparing the organisation and precedes full deployment.
Equipment installation and system configuration
Installation follows the previously approved design. The supplier delivers the AGVs and AMRs and the system components, and infrastructure elements appear on the shop floor one after another. Magnetic tape is laid on the floor or sensors are installed, if the chosen navigation method requires it; with natural navigation, this is the point at which the facility map is loaded into the vehicles' memory. In parallel, the fenders, signs and signal lights provided for in the safety design are installed, and the docking and charging stations are positioned and connected.
The vehicles themselves also need configuring: attachments for handling specific loads and identification markings are fitted to them, and the fleet management software — the "brain" of the system, which assigns tasks and controls traffic — is launched on a server or controller. Once the equipment is on site, the supplier's engineers work with the IT department to connect the AGV system to the ERP, MES or WMS systems according to the agreed interface, triggering test orders and checking the data flow.
This is also the stage at which the operating logic is defined. AGV missions are set up (for example, moving a pallet from point A to point B), along with task assignment rules — whether tasks are triggered automatically by the WMS or manually by an operator — priorities and behaviour when orders queue up. Traffic parameters are configured too: speed limits on individual sections, waiting times, the response to loss of connectivity, and navigation maps. The whole process is rounded off by preliminary small-scale tests of every function: a single test run along each route, checking pallet pick-up, confirming that messages reach the IT system, and testing the emergency stop with an artificial obstacle.
AGV pilot — how to choose the scope
Before you put the whole fleet to work serving production, test the system within a limited scope. A well-chosen AGV pilot covers a slice of the process that a small number of vehicles can handle on their own. In practice this means one production line, one warehouse area or one shift — a section small enough that a possible problem will not stop the whole plant, and representative enough that the conclusions can be carried over to the rest of the shop floor.
The aim of the pilot is to catch problems and confirm that the system works as intended before it is expanded. It is also a learning period for the team: the supplier usually provides its own engineers on site, so it is worth drawing on their knowledge and getting used to operating the system — these are invaluable practical lessons that cannot be learned from documentation.
Which KPIs to monitor during the pilot
A pilot only makes sense if you measure its results. Without hard numbers it is difficult to say whether the system meets its objectives, and even harder to justify a decision to expand it. It is worth tracking a few specific indicators:
- Cycle time — the average time to complete a transport order, from the order to delivery of the load.
- System availability — uptime without failures, showing the real reliability of the fleet.
- Tasks per hour — how many missions the fleet actually completes per unit of time.
- Delivery accuracy — whether the AGV sets loads down exactly in the right places.
- Downtime and its causes — whether production is waiting for a vehicle, and what exactly lies behind each stoppage.
It is also worth separately observing the impact on people's work — for example, whether the warehouse staff at the line spend less time walking to fetch materials. This is one of the effects AGVs are implemented for, and one that is easy to overlook when looking only at vehicle parameters. The key here is the "causes" column for downtime: the mere fact that a vehicle stopped says little; only the reason — a traffic conflict, no space at the receiving station, loss of connectivity — shows what to improve.
Crew feedback and route optimisation after the first runs
Numbers show what is happening; people explain why. Those working in the pilot area — supervisors, warehouse staff, machine operators — should report their observations on an ongoing basis: whether the AGV is blocking traffic, whether the system interface is clear, where inconveniences arise. This is the cheapest source of improvements you have, and it often points to problems that are invisible in the statistics.
The first runs usually reveal traffic conflicts that could not be foreseen on the plan: a vehicle stops in the same place, two AGVs meet on a narrow section, tasks are distributed unevenly. This is a normal part of the pilot — it is exactly why you run one. When you notice a recurring problem, analyse the cause and make corrections. Sometimes a configuration change is enough, for example adjusting a route or a speed parameter; sometimes a minor organisational change, such as arranging loads differently at the receiving station. The pilot is the time to fine-tune the system in real conditions, not in a simulation.
Performance tests and criteria for moving into continuous operation
Before you consider the pilot complete, test the system under load. As part of performance testing, try to reach the target operating level — order the number of missions you expect at peak production hours — and check whether the fleet keeps up. If it does not, that is a signal that an additional vehicle or a change to the task schedule may be needed. It is better to find this out now, in one area, than after full launch across the entire shop floor.
Base the decision to move into continuous operation on evidence, not on the calendar. The pilot can be considered successful when it runs stably for an agreed period — usually a couple of weeks — and all corrective actions have been implemented and verified. A practical condition is that the key indicators stay at an acceptable level without anyone having to rescue the situation manually, and that downtime does not recur for the same reason. Also make sure that, before expansion, all end users have been trained and accept the new solution — without this, even a technically sound system will meet resistance.
Carry out the expansion itself gradually. If two vehicles worked on one shift in the pilot, do not move straight to the full fleet on all shifts — add vehicles and areas in stages, observing how the system copes with denser traffic. Some problems only emerge at scale, and a smooth transition from pilot mode to normal operation protects production continuity.
A success message to the crew
A successful pilot is not just data in a report — it is also an argument that builds acceptance. When the first weeks of operation go well, it is worth showing this to the team in concrete numbers: how many loads the vehicles carried, how high delivery accuracy was, how many monotonous runs people no longer had to make. A message like this boosts morale and defuses concerns more effectively than any reassurances before launch. The sooner the crew sees that automation is taking over tedious transport runs rather than threatening their position, the less resistance you will encounter during full deployment.
AGV pilot step by step — summary
The installation and pilot stage comes down to a simple sequence that is worth keeping in view:
- Install the equipment and infrastructure according to the design, and configure the vehicles and fleet software.
- Integrate the system with ERP/MES/WMS, and define missions, assignment rules and traffic parameters.
- Choose a narrow, representative pilot scope: one line, area or shift, a few vehicles.
- Measure cycle time, availability, tasks per hour, delivery accuracy, and downtime and its causes.
- Gather crew feedback and, after the first runs, correct routes and traffic conflicts.
- Carry out performance tests under the target load.
- Move into continuous operation only after several weeks of stable operation and user training.
- Show the team the pilot results in numbers — build acceptance before full deployment.
A well-run pilot limits risk to a single area and provides hard data for the decision to scale up. It is the cheapest moment to catch errors — and the best moment to convince people with numbers, not promises.
How long should an AGV pilot last?
There is no single fixed figure, but in practice people talk about several weeks of stable operation. More important than the time itself is whether the system maintains the key indicators without anyone having to rescue the situation manually, and whether downtime recurs for the same reason. The pilot ends when evidence — not the calendar — confirms readiness for expansion.
Does a pilot stop production?
It should not. The whole point of starting small is to limit the impact on day-to-day work to a single area or a single shift. Installations are planned to minimise downtime — for example at weekends or during maintenance shutdowns — and the system is expanded in stages. In some deployments it is worth keeping the existing way of working for a while as a safeguard, just in case.
What if the pilot reveals insufficient throughput?
This is one of the most common and most valuable conclusions from a pilot. If the fleet cannot keep up with the target number of missions during performance tests, this usually means an additional vehicle, a change to the task schedule or a route correction to eliminate traffic conflicts is needed. Catching this in a single area is much cheaper than after the system has been launched across the entire shop floor — which is why a well-sized fleet, for example from the AGV All-in-One series, should have spare capacity built in at the design stage.
