Muskeology
Frontier tech, minus the hype

Robotics

Why Robot Maintenance Is The Hidden Cost

Purchase price is a small part of what a deployed robot costs, because calibration, spares, downtime and the people who manage it accumulate for the machine's whole life.

Silhouette of wind turbines on a hill with a dramatic cloudy sky at sunset.
Silhouette of wind turbines on a hill with a dramatic cloudy sky at sunset. · Photo via Pexels

Automation projects are usually justified on the purchase price of the machine. The ongoing costs of keeping it working are larger, less visible and frequently underestimated.

Calibration drifts continuously

Robots are positioned relative to their surroundings by a calibration that assumes fixed geometry between the machine, its tools and the workspace.

Collisions, temperature changes, floor settlement and ordinary wear all shift that geometry slightly, and accuracy degrades until the calibration is redone.

Recalibration takes the cell out of production and often requires a specialist, which is why the interval is a real operational cost rather than a technicality.

Wear items are numerous and unglamorous

Cable harnesses that flex with every motion eventually fail, and they are frequently routed in ways that make replacement time-consuming.

Gearboxes need lubrication on schedule, tool changers wear, and vacuum systems lose suction as filters clog and seals age.

None of these are expensive individually, but together they generate a steady stream of interventions that must be planned and staffed.

Downtime costs more than the repair

When a robot sits at the centre of a process, its failure stops everything downstream, and that lost output usually dwarfs the cost of the part.

This is why spares holding matters. Waiting days for a component turns a short repair into an extended outage.

Operations with several identical machines can hold shared spares economically, which is an advantage smaller deployments cannot match.

Software needs maintenance too

Programs written for a specific product or layout require updating whenever either changes, and those changes arrive from the business rather than from engineering.

Controller firmware and safety configurations must be kept current, and updates require revalidation before the machine returns to production.

Organisations that lack anyone able to modify the programs become dependent on the integrator for changes that ought to be routine.

People are the largest recurring item

Automation reduces the labour needed to perform a task while creating a need for people who can diagnose, adjust and repair the equipment.

Those skills are scarcer and better paid, and a site with only one capable person is exposed whenever that person is unavailable.

Realistic business cases account for this substitution rather than assuming the labour cost simply disappears when the machine arrives. Training existing staff is usually cheaper than recruiting, and it retains knowledge of the process itself.

Service contracts shift some of this burden to the supplier, at a price that reflects the same scarcity, and they work best where response times are specified and enforced.

agricultureharvestingweedingautonomy
Tobias Nkemelu
AI & Compute, Muskeology

Tobias builds and breaks machine learning systems for a living, which makes him a difficult audience for benchmark announcements.

More from Tobias →

Also by Tobias Nkemelu

Robotics

Autonomous mobile robots outside the warehouse

Hospitals, hotels, factories and pavements — where wheeled autonomy has spread, and the specific reasons each environment is harder than a warehouse.

Lena Brandt··3 min read

Neurotech

Reading a neurotechnology paper

What to look for in a study, and which sections tell you whether the headline is supported.

Ravi Shankaran··3 min read