200 graves a night: the unglamorous case for robotics in grounds care

A cemetery nursery lives on long-term grave-care contracts: relatives who no longer live nearby or are no longer able pay for a grave to be planted, tended and watered. Watering is the part that cannot be postponed. In summer it comes every other day, daily in heat, and it comes before the cemetery opens, because nobody wants to work a hose between mourners.

Heino Schwarz, proprietor of Blumen Schwarz in Schwabach, put it to VDI nachrichten this way: watering by hand had become an increasing challenge in the hot summers. That is the whole starting position. No vision, no skills-shortage chart, just a job that finds the fewest hands in the weeks of greatest need, because it starts at five in the morning.

It is also the job a mowing robot does not do. Between rows of graves little is mown and much is carried: water, in cans, along gravel paths. Anyone thinking about robotics on a cemetery therefore has to start with the watering can, not the lawn.

Rainos is a wheeled vehicle with a 215-litre water tank, 165 kilograms empty, 111 centimetres high, with a top speed of 3.2 km/h. Its manufacturer, Innok Robotics of Regenstauf near Regensburg, says it built the machine at the suggestion of cemetery gardeners; over 80 were in use as of December 2023, mainly in the German-speaking countries.

The workflow is simpler than the word autonomous suggests. At initial set-up the manufacturer creates a map of the cemetery. On that map, in a browser, the nursery selects the graves to be watered and sets the water quantity and watering pattern for each. The robot plans the route itself. It drives at night, because nobody is on the paths then, and it is quiet enough that this would hardly matter.

When the tank is empty it drives to a refill point, refills by itself and continues the round. According to the manufacturer it manages up to 200 graves in one night this way; battery life is given as around 14 hours, charging as about three hours a day.

The sentence about the refill point is the one a project hangs on, and reports usually treat it as a subordinate clause. A refill point is a water connection with a docking position that the robot finds at night without help. It has to stand where the round passes, and at intervals that match the tank. Anyone who wants to water 200 graves therefore first plans a network of water points across the cemetery — and that is work for the nursery and the cemetery administration, not for the manufacturer. Much as with charging a robot fleet in a building, the supply infrastructure is the part that has to be finished before the robot.

The trade press quotes a Rainos output of up to 20,000 litres a night. The figure is a manufacturer's upper limit, and it is no use for planning, because a glance at the tank puts it in its place: 20,000 litres from a 215-litre tank is over ninety refill stops in a night. Anyone who actually wants to deliver that volume is planning not a robot but a refill network reachable every few minutes.

The number a nursery plans with is therefore the other one: up to 200 graves a night, at the water quantities stored per grave, with the refill stops the network allows. From that follows how many nights a portfolio needs, and from that whether one robot is enough. The path to the number runs through the map and the layout of the cemetery — not through the datasheet.

This is no criticism of the manufacturer. It is the same pattern as with any machine whose brochure figure is measured under ideal conditions: the useful number is the one tied to supply. What that looks like for drone docks, whose launch time is in the brochure and whose charging cycle sets the day, is described in the article on drone-in-a-box.

The map created at set-up is more than a navigation base. It holds which graves are watered, how much and how — in other words the core of the grave-care contract in machine-readable form. When a contract is added, a grave is activated on the map. When one expires, it is deactivated. When the cemetery administration changes a path, the route changes.

So it should be settled before commissioning what usually only comes up afterwards: who maintains the map, who holds a copy, what happens if the manufacturer can no longer be reached, and whether the nursery can add graves itself or needs a service appointment to do so. That is not a question about Rainos in particular but about any robot whose knowledge of the site lives in a file — the article on robot markers describes the same point for inspection robots in buildings. On a cemetery the file is, in addition, a customer list.

VDI nachrichten gave the orders of magnitude in April 2023: around €80,000 to buy, up to €25,000 a year in saved labour, payback in about three years. The manufacturer puts the break-even at around 215 graves.

The figures are statements by the manufacturer and the operator, not an independent costing, and should be read as such. What can be said independently of them is the structure of the calculation, and it differs from a mowing robot's at one point: a cemetery nursery sells the watering. The grave-care contract has a price, and the watering round is a cost block within it. A robot that takes over the round therefore does not only cut costs; it makes contracts acceptable that the nursery would otherwise have to turn down for lack of staff. The €25,000 describes one side; the other is the number of graves it can take on again.

Water is the third item. The manufacturer states that targeted night watering saves up to 15 percent water compared with watering by day — less evaporation, fixed quantities per grave. On a cemetery where the nursery pays for its water, that is no side effect.

It is worth pausing to note why the watering round suits a robot where mowing on a cemetery mostly does not.

• The round is fixed. The same graves, the same paths, every other day. Nothing has to be recognised; everything is on the map.
• The paths are paths. A cemetery has surfaced paths between the rows, and the machine stays on them. It does not drive over graves or lawn edges.
• The time is night. The cemetery is closed, the paths are empty. The hardest question in service robotics — what happens on meeting people — does not arise, because no meetings happen.
• The product is measurable. A grave is watered or it is not, with the quantity stored. There is no argument about the cut.

Mowing between graves has none of these properties: irregular areas, an edge at every stone, the public by day. That grounds-care robotics is thought about from the golf course outward has therefore done it little good on the cemetery. The case that holds up is the boring one.

Three caveats.

It is one documented site. The Waldfriedhof Schwabach is the case that appears in the trade press with a name, an operator and figures. The manufacturer reports over 80 robots in use, but the other sites are not described in the same way. Anyone checking transferability to their own cemetery checks paths, water points and grave count on the ground, not the reference list.

It is one manufacturer. We know of no second series-built machine for this task. That is convenient for selection and uncomfortable for dependence; the questions of a copy of the map, spare parts and service route therefore belong before the signature, not after. What an operator does when a robot manufacturer fails is set out in the operator's guide to supplier failure.

The performance and economic figures come from the manufacturer and the operator. 200 graves, 14 hours, 15 percent water, €25,000 — none of these has been measured by a third party. They are consistent across several publications, and they are plausible; but they are statements, not findings. Planning should start from your own grave numbers and carry the manufacturer's values as an upper limit.