Before You Buy a Painting Robot
20 questions every decision-maker should ask.
[This checklist is a companion to Before the Robot, Redesign the Work, where I make the broader argument that construction robotics is not only a machine-design problem. It is also a production-system problem: readiness, geometry, workfront continuity, sequencing, quality and verification determine how much value the machine can actually create. That article asks what has to change in painting before robotics can work at its best.]
This article asks the next question: if you are considering buying the robot today, what should you test before signing the cheque?
A robot demonstration can answer an important question:
Can the machine perform the task?
A purchase decision requires answering a much larger one:
Can this machine improve the economics, quality, safety or reliability of my complete painting operation under the conditions in which I actually build?
Before buying the robot, a decision-maker should be able to answer at least the following questions.
1. Where is the actual constraint?
Is painting application itself limiting output—or is the real constraint surface preparation, putty, sanding, readiness, access, drying, inspection, rework or availability of continuous workfront?
A robot that accelerates a non-bottleneck may produce an impressive demonstration without materially accelerating the project.
2. Why does this problem require a robot?
Could the same constraint be removed more cheaply through:
better sequencing
workflow redesign
a power tool
airless spraying
mechanised sanding
material pumping
improved access equipment
better training
a specialised semi-automatic machine?
The correct comparison is not simply robot versus manual labour.
It is:
Robot versus the best practical alternative for solving the same problem.
3. What exactly will count as success?
Define the output before buying the machine.
Is success:
sqft sprayed per hour?
sqft completed per shift?
accepted finished sqft per day?
fewer painter-hours?
lower rework?
better finish consistency?
reduced exposure to hazardous work?
faster project completion?
improved schedule reliability?
For painting, accepted finished output is usually more meaningful than robot motion or instantaneous application speed.
4. Will faster execution actually change project throughput?
If the robot makes one operation three or four times faster, what happens to the overall sequence?
Does the next activity start earlier?
Does possession move earlier?
Does another operation simply become the new bottleneck?
Ask not only:
How fast is the robot?
Ask:
Where does the time saved by the robot actually go?
5. Can enough robot-ready work be supplied continuously?
How many sqft are genuinely available to the robot—not theoretically contained in the project?
Measure ready work by:
location
floor
room
surface condition
access
predecessor completion
time window
A building may contain millions of square feet of paintable area while offering surprisingly little continuous robot-ready area at a particular moment.
6. What is the robot's real operating envelope?
Precisely where can it work?
Test:
large walls
small rooms
ceilings
beams
columns
corners
wall-ceiling junctions
door and window reveals
balconies
utility areas
staircases
double-height areas
façades
irregular geometry
occupied or partially completed spaces
Then calculate what percentage of the actual project lies inside that operating envelope.
7. How much of the building must still be painted manually?
A robot may perform 80% of one surface and still leave:
corners
edges
narrow areas
reveals
masking boundaries
corrections
touch-ups
inaccessible geometry
If a manual crew must follow the robot everywhere, understand whether that crew becomes significantly smaller—or merely changes what it does.
8. What conditions must exist before the robot can start?
Define robot-readiness explicitly.
For example:
required flatness
verticality
dryness/moisture
surface cleanliness
completed putty/sanding
approved material
masking/protection
obstruction-free area
power
lighting
ventilation
safe access
completed preceding trades
The more conditions the machine requires, the more important the system producing those conditions becomes.
9. What happens when those conditions are not met?
Real construction sites produce exceptions.
Can the robot:
detect the exception?
stop safely?
bypass that area?
request human intervention?
resume without complete recalibration?
record why work was skipped?
A machine should be judged not only by how it performs normal work, but also by how gracefully it handles abnormal work.
10. How much movement, setup and changeover does the robot require?
Measure the complete deployment cycle:
Transport → unload → move → position → calibrate → protect/mask → prepare material → execute → inspect → clean → relocate.
Include movement:
between rooms
between flats
through doors
into lifts
between floors
between towers
between projects
The machine's production rate while painting is only one part of its day.
11. Which exact finishing operations can it perform?
Do not buy a “painting robot” as an abstract category.
Determine whether it can handle:
primer
putty
sanding
intermediate coats
emulsion
exterior coatings
texture
specialty coatings
touch-up
inspection
And determine whether changing operations requires different hardware, tooling, software or operators.
12. Is it compatible with the materials actually used on our projects?
Test real products—not merely water or demonstration coatings.
Understand:
viscosity range
particle size
pressure requirements
nozzle limitations
filtration
pot life
settling
material pumping distance
dilution
colour/product changeover
cleaning requirement
wastage
overspray
The robot and the coating system must be evaluated together.
13. Can it repeatedly achieve the required finished quality?
Define quality numerically wherever possible.
Depending on the operation, examine:
coat uniformity
coverage
film thickness where relevant
texture consistency
misses
runs/sags
overspray
edge quality
surface damage
sanding uniformity
visible defects
rejection and rework
The relevant productivity number is not:
area processed.
It is:
area processed and accepted.
14. How does the robot know what it is seeing?
Understand exactly what its sensing establishes.
A camera, LiDAR or scanner that recognises geometry does not necessarily establish:
moisture
adhesion
contamination
surface soundness
coating compatibility
Ask what each sensor measures, at what accuracy, how frequently, and what happens when confidence is low.
For automated inspection, understand both:
false acceptance and false rejection.
15. How autonomous is it—really?
Words such as “autonomous” can hide very different operating models.
Ask how many human-hours are required for:
setup
calibration
supervision
material replenishment
repositioning
obstacle handling
fault recovery
cleaning
inspection
maintenance
One operator supervising several machines is very different economically from one operator continuously attending one robot.
Calculate total human-hours per accepted sqft, not simply the number of people touching the applicator.
16. What happens when the robot fails?
Construction equipment operates in dust, heat, moisture and changing site conditions.
Understand:
expected uptime
failure modes
recovery time
remote support
availability of technicians
spare parts
consumables
preventive maintenance
calibration
battery life where applicable
software dependence
internet dependence
A fast robot that regularly waits days for specialist service may have poor real productivity.
17. What new safety and environmental risks does it introduce?
Automation may remove workers from hazardous tasks, but it can create new risks.
For painting, assess:
high-pressure spray
airborne coatings
overspray
ventilation
electrical equipment
moving machinery
collision
working at height
battery hazards
hoses/cables
emergency stopping
interaction with other workers
The comparison should be:
risk before automation versus risk after automation, not simply “robot equals safer.”
18. What is the full economic cost—not merely the purchase price?
Calculate the complete deployed cost:
machine price
financing
import/tax/shipping if applicable
software subscription
licences
attachments
pumps/compressors
access equipment
batteries
consumables
operators
support labour
training
maintenance
spares
transport
insurance
storage
cleaning
mobilisation
downtime
expected economic life
residual value
Then compare it with an optimised manual or machine-assisted baseline, not an inefficient traditional crew.
19. What utilisation is required for the economics to work?
Build the financial model backwards.
If the robot must operate productively for, say, a certain number of hours or sqft every month to justify its cost:
Can our actual projects reliably provide that workload?
Stress-test the economics against:
lower utilisation
fragmented workfront
project gaps
slower mobilization
support labour
maintenance
lower-than-promised productivity
Indian labour rates
shorter equipment life
The economics of construction robotics may be determined as much by the hours in which the robot is not producing as by how fast it works when it is.
20. What evidence would make us comfortable scaling from one robot to ten?
Before a large purchase, define the pilot pass/fail criteria.
Require evidence under representative site conditions for:
accepted output
quality
human-hours
utilisation
downtime
setup/changeover
material consumption
rework
safety
total cost
Also examine the vendor:
How many robots are actually operating on construction projects?
For how long?
Under what conditions?
Can existing customers be contacted?
What warranty and service commitments exist?
Who owns operational data?
What happens if the supplier fails?
Can the equipment continue working if software support ends?
A successful demonstration answers:
“Can the robot do it?”
A successful pilot must answer:
“Can our production system use it repeatedly enough to justify owning it?”
The final purchasing question may therefore be the simplest:
Am I buying a faster machine—or am I building a production system capable of keeping that machine productive?
The distinction determines whether the robot becomes a productivity asset or an expensive piece of equipment waiting for work.
Perhaps the deeper realization is that roughly half of these questions are not really questions about the robot at all. They are questions about the surrounding production system—readiness, workfront continuity, geometry, logistics, quality definition, sequencing, data and exception handling.
The more questions you ask before buying the robot, the more you discover that robotics in construction is partly a robot problem—and largely a production-system problem.