AMR Fleet Charging: The Fleet Stops Before the Battery Does

Anyone running a shift notices it at the same moment every time: mid-afternoon, missions queuing, and two vehicles parked beside the charging station. Neither is truly empty, but the third one to arrive finds the spot taken and waits, and the pallet round that was meant to close before shift change slips. The question the supervisor then takes to the supplier is usually about battery runtime. It is almost always the wrong question. Fleet availability is settled by the geometry of the dwells and by the number of charging points, long before it is settled by the watt-hours in the pack.

Three Numbers, and the Rest Is Arithmetic

Sizing starts from energy rather than power, and it needs three quantities the site already owns: usable pack capacity, the energy a typical mission consumes, and the minutes genuinely available at a station during a peak shift. Average power required is simply the energy to be replaced divided by the window in which you can replace it, and from there the arithmetic does its own talking. An in-process charging point rated 3 kW returns 0.3 kWh in a real six-minute dwell. If the average round consumes 0.25 kWh, the balance closes each cycle. If it consumes 0.6, the fleet gives back half its need every round, and the drift stays invisible until the state of charge drops through the floor in the worst hour of the day.

The second piece of arithmetic is saturation, and it dismantles most of the optimistic projects. A vehicle running at 95% saturation across an eight-hour shift has 24 minutes of standstill in total, broken into one- and two-minute fragments wherever they fall. No charging strategy survives that distribution, however fast the charger. Adding a vehicle lowers saturation for everyone and creates the window; it costs less than it looks, because the extra machine isn't there to move more pallets but to let the others charge. For in-process charging points, the orders of magnitude suppliers quote sit between 3 and 6 kW per point, and that is a figure worth checking against your own switchboard before checking it against a catalogue.

Three Strategies, Three Ways to Lose Vehicles

There are only a few workable strategies, and the choice follows the gaps in the cycle. Opportunity charging uses the natural dwells, many and short, and keeps vehicles in rotation. Scheduled charging concentrates the energy in one long session, which is simpler to manage but removes the vehicle from the round for a block of time. Pack swapping cuts downtime to minutes, at the price of spare packs, dedicated space and a manual procedure somebody has to perform. The first model rewards chemistries that tolerate frequent partial charges, while old lead-acid pays for them through sulfation, as a review published on Robotics Tomorrow in August 2026 by an intralogistics integrator points out.

The same text carries the methodological advice that outweighs any single figure: before choosing, simulate the state of charge of the whole fleet across a peak shift, at the real duty cycle rather than the average one. Sizing done on the average works eleven months of the year and fails in the month that matters.

The Queue Grows Faster Than the Fleet

That the bottleneck is the station rather than the cell is visible in published data. In a real-warehouse study in Frontiers in Neurorobotics in 2023, Hu, He and Shen compared fleets of 5, 10 and 15 AGVs on a layout with five charging stations: delay times rise from 219 seconds with 50 missions and 5 vehicles to 3,591 seconds with 100 missions and 15 vehicles, and hierarchical planning that handles missions and charging together cuts average completion time by 13.62% and delays by 76.69% against the sequential approach. The two operating constraints the authors model are the ones every supervisor knows: a vehicle has to finish its current mission before going to charge, and it cannot turn up at the charger with a load on board.

On threshold behaviour, the most recent reference is a reinforcement learning simulation published by Shaji, Sobhanan and Defryn in July 2026, on scenarios with four and six vehicles and only two stations. The learned policy sends vehicles to charge when state of charge falls between 24% and 28% and returns them to service between 75% and 95%, completing about 6% more orders than the best benchmark heuristic. The study's units are abstract and the values don't transfer directly to a site, but the shape of the policy does: you leave the charger well short of 100%, because the last few percentage points cost station time and take it from the vehicle that is waiting.

Who Decides When a Vehicle Leaves the Round

Not the robot. The fleet manager decides, and that is where to look when the fleet mixes brands. VDA 5050, which reached version 3.0.0 in March 2026, standardises exactly that interface over MQTT with JSON messages, and for charging it defines the data and the commands: states batteryCharge, batteryHealth, charging and reach, and the actions startCharging and stopCharging, available both at a charging spot with the vehicle stationary and on a charging lane travelled in motion. Safety, traffic management and cybersecurity stay outside the scope, while energy management is listed among fleet control functions. The specification makes charging uniformly observable and commandable, and leaves the fleet owner to decide how many stations are needed and at which thresholds to use them.

One detail of that same specification deserves attention at purchase time. The field declaring battery health is a percentage, but the criterion behind the percentage lives in the manufacturer's BMS, not in the standard: two fleets both reporting 85% can be in very different condition. Anyone assembling vehicles from different suppliers should have each of them state, contractually, how that number is defined and at which threshold the pack counts as end of life.

Chemistry Decides How Often You Can Break the Charge Up

For the two common chemistries the reference figures, taken at cell level rather than pack level, remain those summarised by Battery University: lithium iron phosphate around 90-120 Wh/kg, with typical 1C charging, more than 2000 cycles and a thermal runaway onset near 270 °C; NMC rising to 150-220 Wh/kg with 1000-2000 cycles and onset nearer 210 °C. These are category ranges, useful as orders of magnitude rather than as product specifications, and they are enough to steer the choice. Where the vehicle can carry the weight, cycle life and thermal margin pay off in a fleet taking dozens of partial charges a day. Where the pack has to fit a compact chassis, energy density becomes the deciding argument.

On the hidden cost of hurry there is a quantitative figure, although it comes from simulation. In the battery-health-aware dispatch study published on arXiv in 2026, the model gives fast charging roughly three times the wear of standard charging and adds calendar ageing for vehicles left idle at high state of charge; a policy accounting for both cuts total degradation by more than half against rule-based dispatch, at equal throughput. The coefficients are modelling proxies rather than cell measurements and should be read that way, but they show where the time won with a bigger charger ends up: in the life of the pack.

Below Zero the Question Isn't How Much Runtime Is Left

In refrigerated warehouses the constraint changes nature. It isn't about how long the battery lasts in the cold, it is about whether charging is permitted at all: below 0 °C lithium tends to plate out as metal on the anode instead of intercalating into the graphite, which costs capacity permanently and leaves the cell vulnerable to vibration afterwards. Battery University is blunt about the practical limits: no charging below freezing, and at -30 °C the admissible current drops to a fiftieth of nominal capacity, which puts a full charge in the tens of hours; some special cells accept current down to -10 °C at reduced rate and lower end-of-charge voltage, reaching roughly 80% of capacity in more than twelve hours.

Chen and colleagues described the mechanism quantitatively in a 2025 study in ACS Electrochemistry: plating begins once polarisation drives the anode potential below 0 V versus lithium metal, and the capacity accessible before that point collapses from 412 mAh/g at -20 °C to 160 at -30 °C and 106 at -40 °C. Those are measurements on a graphite electrode in a half cell rather than on an industrial pack, so they carry as mechanism rather than as system performance; the operational consequence is the same either way. In cold storage there are two roads: put the station outside the refrigerated zone and budget the transfer time, or adopt packs with integrated heating managed by the BMS that bring cells up to temperature before accepting current. For comparison, lead-acid pays the cold in capacity rather than in prohibitions: a technical manager at a truck manufacturer, writing in a trade magazine in December 2024, put the loss near 30% of nominal capacity below freezing, and that is an interested party's statement, good as an order of magnitude.

The Charging Area Stays a Technical Room, Lithium or Not

The standards split into different families, and mixing them up produces quotations that forget something. Machine safety is governed by EN ISO 3691-4:2023, covering driverless industrial trucks and their systems, charging stations included. The truck's electrical requirements sit in EN 1175:2020, which includes lithium energy sources and the functions used in the automatic operation of driverless trucks, and whose new edition was approved by CEN in May 2025 and should be checked for harmonisation status and coexistence period before being quoted in a specification. Safety of industrial lithium cells and batteries, motive applications of industrial trucks included, is in IEC 62619:2022, whose second edition added requirements on moving parts, hazardous live parts, battery system design and electromagnetic compatibility. Installation and operation of traction batteries, explicitly including those of automated guided vehicles, is in IEC 62485-3:2014, which carries the method for calculating ventilation during charging.

That ventilation method was born from lead-acid gassing, which is why in many plants the charging area is a ventilated room with a sign on the door. Moving to lithium doesn't remove the risk, it changes its nature: no longer hydrogen to dilute but thermal energy to contain and separate, with the same design seriousness applied to a different problem. That reading is ours rather than a normative citation, but it is the one that prevents the worst conversation, the one where charging gets installed wherever there happened to be room.

The Number Quoted, and the Figure That Eats It

On the benefits of in-process charging, the most quoted European figure comes from a September 2024 consultancy study circulated by the supplier of the inductive charging system: "+50% productivity" for the AGV fleet, "-30%" fleet size in an alternative scenario, "more than 5,000 hours a year" of charging downtime avoided, "10,900 crossings avoided" between people and vehicles. The conflict of interest is declared by the source itself, fleet size and shift count are not published, and without those two numbers none of the percentages can be verified. Next to that 50% belongs the figure that erodes it, the same one seen above: the strategy that keeps vehicles always topped up is also the one that charges them fast, and fast charging costs pack life by a margin that is anything but negligible.

The third-order effect is the sneakiest, and it involves people more than machines. When the station sits off the natural round, vehicles go there reluctantly and shift leaders learn to postpone until the state of charge turns critical; at that point nearly all of them go to charge together, in the worst possible moment, and the queue you were avoiding forms anyway. Station placement, in other words, isn't a plant engineering choice. It is an organisational choice dressed as one.

How a Serious Feasibility Study Is Set Up

Credible sizing doesn't start from a battery catalogue. It starts from warehouse data the site already holds, even when nobody has lined it up before:

• missions per shift and their hourly distribution, with the peak hour kept separate from the average;
• average distance per mission, loads handled, and any ramps or gradients;
• number of shifts per day and length of breaks, which are the candidate windows;
• a layout showing where vehicles already dwell today, and the electrical capacity available there;
• operating temperatures, refrigerated zones in particular;
• expected vehicle saturation, because above 90% the windows disappear.

Out of those numbers come pack capacity, the number and position of stations, the charge entry and exit thresholds and, more often than not, the real recommendation: if the average dwell window returns less energy than the round consumes, no battery fixes it, and the choice is between moving the station onto the route, adding one, or adding a vehicle. PITECH sets up this evaluation together with the people running the warehouse, within its material handling automation and mobile robotics solutions and, for pallet movement, its robotic forklifts and AMRs. The wider business case, with the items that genuinely belong in the return calculation, is covered in the guide on AMR ROI, TCO and payback, while phasing robots into a warehouse that never stops is the subject of the guide on automating pallet handling in a running warehouse. To start the study, one peak shift of data is enough, from the contact page.