Construction robotics and jobsite automation field guide

Construction robotics, and why it augments the crew

Construction robotics is the use of robots and automated machines to take on the dull, dirty, dangerous, and repetitive work on a jobsite. A robot prints the layout from the model onto the slab. A machine-control dozer cuts earth to grade by GPS. A remote machine breaks up a structure with no operator inside it. A ceiling robot drills the overhead anchor holes off the model. None of these invent anything. They run a structured, repeating task faster and more consistently than a crew does it by hand, and they do it without putting a person in the spot that hurts people.

The framing that matters more than any spec sheet: these machines augment the crew, they do not replace the trade. The robot does the part that is repetition and strain. The skilled trade does the judgment, the setup, the exceptions, and the work that is not structured enough for a machine. A layout robot is dead weight without a coordinated model and survey control behind it, and somebody has to build that, check it, and supervise the print.

That last point is the thread through every category here. The robot is only as good as the model and the control it builds from, and a human supervises it and handles what goes sideways. The model and control side of the work lives in the construction layout guide and the BIM coordination guide. This guide is about the machines that build off them, where they pay, and how to run them without anyone getting hurt.

Robots augment the crew, they do not replace the trade

The fear gets the headline and the reality gets the work done. A construction robot takes a narrow, repeating task off a person's plate. It does not show up with the judgment of a journeyman, the read of a foreman, or the ability to improvise when the slab is not where the drawing said it would be. The trade press and the people actually deploying these machines say the same thing: robots take the three Ds, dull, dirty, and dangerous, plus the fourth that pays the bills, repetitive.

Think about what that frees. The crew that would have spent two days on hands and knees tying ten thousand rebar intersections is now setting the hard junctions, handling the splices, and dealing with the parts the machine cannot reach. The layout crew that would have chalked a whole floor plate by hand is now verifying the print and chasing the conditions the model did not catch. The robot did the repetition. The trade did the thinking. That is the partnership, and it is why the right word is augment.

Be blunt about the limit too. Most of these machines are task-specific. They do one structured thing well and they need a person to set them up, feed them, watch them, and take over when the task stops being structured. A robot that ran flawlessly on a clean deck is useless on a congested one until a human sorts the congestion. The skilled trade is not going away. The grind is.

Why robots on the jobsite now

Four pressures pushed construction robotics from pilot to production, and labor leads. The workforce is aging faster than it is being replaced, and a shrinking pool of skilled hands makes the math on a machine that runs a repetitive task far easier than it was a decade ago. When you cannot hire the crew, a robot that does the grind starts to look like the only way to keep the schedule.

Safety is the second pressure, and it is the cleanest argument. A remote demolition machine keeps the operator out of an unstable structure. An autonomous compactor keeps a person off a machine running the same lane all day. A ceiling robot takes a worker off the lift doing overhead drilling, which is where shoulder and neck injuries pile up. Removing the person from the dangerous and the ergonomically punishing task is the win that needs no spreadsheet, though the manufacturer's safety case and the applicable standard still govern how you run it.

Consistency and productivity round it out. A machine that ties or drills or grades the same way every time does not get tired at hour seven, and it logs what it did. On the repetitive, high-volume task, that consistency and speed is where the productivity case is real. On the one-off, it is not. Hold those two pressures, safety and the repetitive volume, separate from the hype, because they are the ones that hold up on a job.

The categories of construction robot

Construction robots sort by the task they take, not by how impressive they look in a video. Each category below maps to a repetitive or dangerous job that a crew used to do entirely by hand. The named examples are real machines in the field as of this review. Treat the capability and accuracy figures as the manufacturer's claims to verify against your own conditions, not as guarantees.

The pattern across all of them is the same. The robot runs a structured task off a model or a control reference, and a human sets it up and supervises. Match the category to a task you actually repeat at volume, and the rest of this guide tells you what model and control it needs and how to run it safely.

Layout robots: printing the model on the slab

A layout robot takes the coordinated model and prints the building layout, full size and one to one, onto the slab or deck. Wall lines, openings, anchor locations, sleeve and embed marks, all the points and lines a crew used to chalk by hand come off the model and onto the floor in the trade's own colors and labels. FieldPrinter from Dusty Robotics and HP SitePrint are the two best known. The pitch is speed and fewer transcription errors, because the marks come straight from the model rather than from a tape and a person reading a dimension.

The accuracy and the positioning method differ by machine, and that is where you read the manufacturer's spec carefully. FieldPrinter cites placement to about 1/16 in and prints without line of sight, including past obstructions. SitePrint cites about 1/8 in and stakes out from a robotic total station, so it needs an unblocked view. Both put the model on the ground far faster than a two-person chalk crew, and both are only as good as the model and the control they print from.

This is the cleanest example of the whole guide's thesis, so it is the right one to start a program on. A layout robot does not replace the layout trade. It replaces the kneeling and the chalk. The field engineer still owns the control, verifies the print against it, and handles the conditions the model missed. How that model and that control get built, and the choice between a total station and GPS behind it, is the construction layout guide. The coordinated model the robot prints from is the BIM coordination guide. Garbage in either one and the robot prints the error perfectly.

Autonomous and machine-control equipment

The earthwork side of automation is the most mature, because GPS machine control has been on dozers and graders for twenty years. A 3D machine-control kit, Trimble and similar, takes the design surface from the model and guides the blade to grade automatically, so the operator stops chasing stakes and the cut hits the model. That is semi-autonomous: a person is in the cab, the machine holds grade. It is the productivity floor for site work and it sets the model dependency that the fully autonomous machines inherit.

Full autonomy is moving up the same ladder. Built Robotics fits an autonomous system to excavators for trenching, combining RTK GPS, inertial sensors, and laser rangefinders to localize and dig to plan with no one in the cab. Caterpillar has shown autonomous excavators, dozers, and compactors, and the autonomous compactor is a clean fit because compaction is the same lane run over and over to a pass count, which is exactly the dull and repetitive work a machine should own.

Trimble frames this honestly as a progression, machine guidance first and full autonomy later, not a switch you flip. That framing is the right one to carry. Grade-to-model and pass-count compaction pay because they are repetitive and structured. The judgment calls, the soft spot the operator feels, the utility nobody marked, the change the super made this morning, still belong to people. The machine holds the grade. The crew runs the site.

Demolition robots: keeping people out of the structure

A demolition robot is a remote-controlled, tracked machine with a breaker, shear, or bucket on an arm, run by an operator standing well back from the work. The category leader people name is Brokk. The entire value is one sentence: it keeps the operator out of the unstable, contaminated, or confined structure while the structure comes down. That is the dangerous D, handled directly.

Think about where this pays. Interior demolition in a building of unknown condition, a fire-damaged or partially collapsed structure, work near contamination, hot or confined spaces where you do not want a person breathing the dust or standing under the load. A remote machine has reach and breaking power past what a person with a handheld breaker can manage, and the operator watches from a safe distance with a clear view. Nobody is under the slab when it drops.

It is still a powerful moving machine with a heavy tool swinging on an arm, so the work zone and the exclusion around it are not optional. Being remote moves the operator to safety. It does not make the machine safe to stand next to. Follow the manufacturer's operating envelope and the demolition safety plan, and keep the zone clear of everyone, including the operator's own feet.

Drilling and MEP robots: overhead off the model

Overhead drilling is one of the worst repetitive jobs in the building: hundreds or thousands of anchor holes in a concrete ceiling, drilled from a lift with arms over the head, day after day, and it wears out shoulders and necks. A drilling robot takes that job. The Hilti Jaibot is the example most contractors have seen: a semi-autonomous mobile platform that drills overhead anchor holes from the model and marks each hole by trade.

The workflow shows the model and control dependency plainly. The hole locations come from a BIM model, exported from Revit or AutoCAD to the cloud, and the robot locates itself on site against a robotic total station for control. It drills dust-controlled and marks the holes for the trades that follow. Hilti reports it drills and measures meaningfully faster than the manual method and cuts the layout-and-drill time substantially. Read those as the manufacturer's figures, gathered under their conditions, and verify the rate on your own deck.

The robot does not replace the MEP trade. It drills the repetitive holes off the coordinated model so the installers spend their time hanging, not drilling overhead. The model has to be coordinated and the control has to be right, or the robot drills a clean grid of holes in the wrong places. The coordination behind those hole locations is the BIM coordination guide, and the control behind the robot's position is the construction layout guide.

Rebar-tying and masonry robots

Rebar tying and bricklaying are both pure repetition with a heavy strain tax, which is exactly the work a robot should take. On the rebar side, TyBot from Advanced Construction Robotics is a gantry that crawls a rebar mat, uses computer vision to find the intersections, positions a tie head over each one, and ties it. The company cites rates above a thousand ties an hour and several times faster than a crew, and a companion machine, IronBOT, handles placing and distributing the heavy bar. On bridge decks and large mats, that is a lot of kneeling and wrist strain removed.

On the masonry side, SAM, the Semi-Automated Mason from Construction Robotics, placed brick in sequence to assist a human mason rather than working alone; the company has since shifted its focus to the MULE lift-assist for heavy material handling, but SAM remains the clearest illustration of the assist model. The name is the honest part. It is semi-automated. The robot does the repetitive pick, mortar, and place, and the mason handles the corners, the cuts, the quality, and everything the machine is not built for. That division is the model for this whole category.

Neither machine ties or lays everything. They take the high-count, repetitive field of the work and leave the judgment, the edges, and the exceptions to the trade. The case is strongest where the volume is high and the pattern is regular, a long deck or a long straight wall. On a small or irregular job, the setup never pays back. That is the ROI line you hold for any of these.

Exoskeletons: augmenting the worker directly

An exoskeleton is the one category that does not replace the task at all. It augments the worker doing it. A wearable frame, usually passive, supports the back, shoulders, or arms and offloads part of the strain from repetitive lifting or sustained overhead work. There is no robot doing the job. There is a person doing the job with less load on the joints that wear out first.

The evidence is on the strain, not the speed. Studies on passive shoulder exoskeletons report large reductions in shoulder muscle activity during overhead work, and field deployments have reported meaningful drops in strain and sprain injuries among workers wearing them. For overhead drilling, ceiling work, and repetitive lifting, that is the same injury problem the drilling robot attacks, addressed from the human side instead.

Treat the exoskeleton as PPE-adjacent equipment with its own fit and training, not a gadget you hand out. The right device depends on the task: a shoulder support for overhead, a back support for lifting, and the wrong one helps nothing. Industrial exoskeletons are covered by emerging standards, ASTM F48 for the devices and ISO 13482 for wearable assist robots, and NIOSH has flagged that the long-term data is still building. Pick the device to the task and follow the manufacturer's fit and use guidance.

Material transport and site rovers

Two more categories round out what is actually in the field. Material-transport robots and autonomous carriers move material across a site or a floor so people are not pushing carts all shift, and they are the least mature of the bunch on most jobs. The bigger presence right now is the site rover for data, and the one everyone has seen is the Boston Dynamics Spot.

Spot is a quadruped that walks a programmed route, the same path every day or week, and captures reality: 3D scans, 360 imagery, and thermal data, with no person walking the hazardous areas. The output feeds progress monitoring and a compare against the model, so the team sees work-in-place against the plan without a person doing the documentation by hand. It is less a builder than a data collector, and that is the point. It does the repetitive walk-and-capture and keeps people out of areas you would rather not send them.

The thread holds here too. The rover captures against the model and the same control, and a human reviews what it brought back and decides what it means. The robot gathers. The team judges. That data loop is where these machines connect to the rest of the field workflow, covered later in this guide.

The common thread: the model and the control

Strip away the differences between a layout printer, a dozer, a drilling robot, and a rover, and the same requirement sits underneath all of them. Every one runs off a model and a control reference. The layout robot prints from the coordinated model and locates itself against survey control. The dozer cuts to a design surface from the model by GPS. The drilling robot pulls hole locations from the BIM model and positions against a total station. The rover compares its scans to the model. No model and no control, no robot.

Say the consequence plainly, because it is the single most important line in this guide: the robot is only as good as the model and the control it builds from. Garbage model, garbage robot output. A robot does not catch a coordination error the way a sharp foreman does. It executes what it was given, perfectly and fast, including the mistake. Feed it a model with a clash nobody resolved or a control point that drifted, and it will print, drill, or grade that error across the whole floor before anyone notices.

So the digital chain is the real work, and it lives in the two sibling guides. The coordinated, clash-free model comes out of the process in the BIM coordination guide. The survey control the robot positions against comes out of the construction layout guide. Get those two right and the robot is a force multiplier. Get either wrong and the robot multiplies the error just as efficiently.

Model and control: the dependency in practice

On the ground, the model-and-control dependency is a setup discipline, not a one-time thing. Before the robot runs, the model has to be the current, coordinated version at the right level of detail, and the control on site has to be established, checked, and matched to the model's coordinate system. A shared coordinate system between the model and the field is what lets the robot land its work where the designer intended. Mismatch the datum or the coordinate system and the robot is precise about the wrong location.

Accuracy is a chain, and the robot is the last link, not the only one. The control network has a tolerance, the robot's own positioning has a tolerance, and the work has a tolerance it must hold. Match the robot's accuracy to what the task actually needs, per the manufacturer, and confirm the control is tighter than the robot so the control is not the thing limiting you. A 1/16 in printer means nothing on control that wandered a quarter inch.

Budget the setup honestly. Establishing and checking control, loading and verifying the model, calibrating the machine, and confirming it on a known point all take time before a single productive unit gets done. That setup time is part of the ROI math, and it is the part the demo never shows you. A trained operator who can verify the robot against control, rather than trusting the screen, is the difference between a machine that helps and a machine that confidently builds the wrong thing.

The human supervises and handles the exceptions

Every machine in this guide is a human-in-the-loop system, and that is by design, not by limitation. The robot does the structured, repetitive part. The person does the setup, the supervision, the quality check, and the exception, which is every situation the robot was not built for. On a real jobsite the exceptions are constant: the congested area the layout robot cannot reach, the rebar junction the tie head misjudged, the soft spot the autonomous machine cannot feel, the condition the model never showed.

This is why the operator is not obsolete, just doing different work. The skill shifts from doing the repetitive task to running and verifying the machine that does it. The layout robot still needs a field engineer who owns the control and checks the print. The drilling robot still needs someone who confirms the model and watches the work. The autonomous dozer still needs an operator's read of the site. The judgment did not leave. It moved up a level.

Treat the relationship as a partnership and the program works. Treat the robot as a replacement for the person and it fails, usually expensively, the first time it meets a condition it cannot handle and there is nobody set up to take over. The robot handles the volume. The human handles the surprises. Build the workflow so the person is always positioned to step in, because they will need to.

Robots are tools, not magic

The single biggest reason robot programs disappoint is the expectation, not the machine. A construction robot is a tool with a narrow job, not an autonomous crew member. It has setup time, it needs maintenance, it needs an operator with real skill, and it only works on a structured task. Walk in expecting that and the machine delivers. Walk in expecting a robot that just builds the building and you will be disappointed by month two.

Be specific about the limits. The machine is built for structured, repeating work with a clear pattern: a regular floor plate, a long deck, a grading surface, a wall of consistent holes. The moment the task gets custom, irregular, or unpredictable, the robot's productivity falls off a cliff and the human has to take over. It breaks down and needs parts and a technician. It needs charging or fuel and a place to stage. It is, in every practical sense, another piece of equipment on the fleet with an uptime number and a maintenance line.

None of this is a reason to avoid them. It is the reason to match the robot to the right task and resource it like real equipment. The contractors who do well with robots are the ones who treated them as tools from day one: picked the structured high-volume task, trained the operator, budgeted the setup and maintenance, and measured the result. The ones who chased the magic spent money and got a robot parked in the trailer.

Where the ROI actually is

The return on a construction robot is built on one thing: repetitive, high-volume work. The machine carries a real cost, the purchase or rental, the setup time, the operator, and the maintenance, and it earns that back by doing a repeating task faster and more consistently across enough volume to matter. Layout on a huge slab. Grading a big site. Tying a long bridge deck. Drilling a thousand-hole ceiling. That is where the per-unit savings stack up into a number that pays for the machine.

The flip side is just as firm. On a one-off, a small job, or a custom condition, the setup never amortizes and the robot loses to a crew. The time to establish control, load the model, calibrate, and verify is roughly fixed whether the task is large or small, so on a small task that fixed setup swamps any per-unit gain. A layout robot does not pay to chalk one room. It pays to print a tower's worth of floors.

So run the ROI on the volume, not on the technology. Count the units, the rate, the setup time, and the rework, and compare honestly against the crew doing it by hand. Where the volume is high and the task is structured, the case is usually clear. Where it is not, no amount of capability on the spec sheet changes the answer. Hedge any vendor ROI claim against your own task volume before you sign.

Where construction robots fit, and where they do not

Pick the task before you pick the robot, because the task decides whether any of this works. Robots fit four conditions, and the more of them a task hits, the better the fit. The work is repetitive and high in volume. The work is dangerous or punishing to a person's body. The work is structured and predictable enough for a machine to follow. And the work is labor-constrained, where you cannot get the crew to do it anyway.

They do not fit the opposite. Custom, one-off work where every piece is different. Unstructured conditions a machine cannot read. Tight, congested, or irregular spaces where setup costs more than the work saves. Low-volume tasks where the fixed setup never pays back. A robot dropped onto a job like that becomes an expensive way to do slowly what a crew would have finished by lunch.

The practical screen is short. Is the task repetitive and high volume? Is it structured? Is it dangerous or strain-heavy? Can you feed it a good model and good control? If the answers are yes, run a pilot. If the task is a custom one-off in a congested space with a shaky model, the answer is a crew, and that is not a failure of the technology. It is matching the tool to the work, which is the whole job.

The robot is a moving hazard

Here is the part that does not get hedged. A construction robot is a moving machine with mass, power, and often a tool, and it is a new hazard on the site the day it arrives. A layout robot drives the deck. A drilling robot moves and drills overhead. A demolition machine swings a breaker. An autonomous dozer cuts grade with no one in the cab. Every one of them can injure or kill a person who is in the wrong place, and the autonomous machines are more dangerous precisely because no operator is sitting there to see the person and stop.

Manage it like the hazard it is. Establish a defined work zone and a people-exclusion around the machine, and keep everyone out of it while the machine runs. Verify the emergency stop before every shift and make sure the operator and the crew know where every e-stop is. For autonomous equipment the zone is the whole operating envelope, not just the spot the machine is in right now, because it moves on its own. Awareness training for the whole crew matters, because the person who walks into the zone is rarely the operator.

The standards exist and they govern, so use them. Autonomous and semi-autonomous earth-moving machinery falls under ISO 17757, which covers stop systems, warning devices, positioning, and zone control at the system level, and remote-control machines fall under ISO 15817. Industrial mobile robots have their own safety standard, and OSHA's general duty and equipment rules still apply on top. Follow the manufacturer's operating envelope and the applicable safety standard, and never let the novelty of the machine soften the rule: people stay out of the zone.

Integrating the robot into the workflow

A robot does not run in a vacuum. It has to fit the sequence, the schedule, and the other trades, and the site has to be ready for it. A layout robot needs the slab clear, clean, and accessible. A drilling robot needs the deck ready and the area clear of the trades it works around. An autonomous machine needs its zone and its control set up. Drop a robot into a job that is not staged for it and the machine waits while a person does the work the robot was supposed to save.

Plan its slot in the schedule like you would any specialty operation. When does it run, what has to be done before it, what does it hand off to, and how does its zone interact with the trades working nearby that day. The hole-drilling robot's output is the layout for the trades that hang off those holes, so its timing ties to theirs. The layout robot's print is the start line for framing and MEP, so it has to land before they need it, not after.

Site readiness is the quiet failure point. The model is current, the control is set, the area is clear, power or charging is available, and the operator is scheduled. Miss one and the robot's productivity evaporates into standby time, which is the most expensive thing a piece of equipment can do. Integration is planning, and it is the part that turns a capable machine into actual progress on the schedule.

The robot reports back

A connected robot does more than the physical task. It records what it did, and that data is half the value once a program matures. The layout robot logs what it printed. The machine-control dozer records the grade it cut. The drilling robot logs the holes against the model. The rover captures progress and as-built against the plan. That stream is progress, QA evidence, productivity numbers, and an as-built record, generated as a byproduct of the work.

Used well, the data closes a loop. The as-built feeds back to the model so the record matches what got built. The QA data shows the work was placed to tolerance. The productivity log tells you the real rate, which is the input you need for the next ROI decision and the next bid. A robot that builds and reports is worth more than one that only builds, because the reporting is what proves the work and improves the plan.

The catch is that the data has to land somewhere a person can use it, not stay locked in a vendor portal nobody opens. Tie the robot's output into the field workflow and the project record so the progress, the QA, and the as-built reach the people running the job. A field platform such as FieldOS is where that output becomes part of the daily record rather than a file on a laptop. Capture it, route it, and the robot pays twice.

How to adopt: start with one task

The adoption pattern that works is crawl, walk, run, and it starts with one task. Pick a single repetitive, high-volume job where a robot clearly fits, the kind that hits the conditions from the where-it-fits section, and prove it on that one task before you do anything else. Trying to robotize the whole job at once is how programs collapse under their own complexity and a frustrated crew.

Run it as a real pilot. Set a measurable goal, the rate, the rework, the hours saved, and compare it honestly against the crew doing the same task by hand. Train the operator properly, because an undertrained operator makes the machine look worse than it is. Budget the setup and the learning curve, because the first run is always slower than the tenth. Then look at the numbers and decide whether it earned the next task.

Scale only on proof. Once the pilot shows the ROI and the crew has the safety and the operating routine down, take it to the next task and the next. The contractors who succeed treat the first machine as a controlled experiment and let the results drive the expansion. The ones who buy a fleet on a sales pitch and deploy everywhere at once usually end up with idle machines and a workforce that does not trust them. One task, proven, then grow.

The workforce: the trade evolves

Robots change the work, and the people side decides whether a program sticks. The new role is the robot operator, the person who sets the machine up, runs it, verifies it against control, and takes over on the exceptions. That is a skilled job, and it usually comes from upskilling the existing crew rather than hiring around them. The hand who knows the layout work is the right person to run the layout robot, because they know when the print is wrong.

Frame it to the crew honestly. The trade is not disappearing, it is evolving, and the repetitive grind is the part going away first. A worker who spent years tying rebar or drilling ceilings is not out of a job. They are running the machine that does it and handling the work that needs a person. That framing, plus real training and a path to the new role, is what gets a crew to adopt a robot instead of resenting it.

Resentment is a real failure mode and it is avoidable. A robot introduced as a threat gets quietly sabotaged or set aside. A robot introduced as a tool that takes the worst work off the crew, with the crew trained to run it, gets used. Invest in the people side as much as the machine, because the best robot on the site does nothing if the crew will not run it.

What to document

A robot program lives or dies on its records, and not just the as-built. You need to know which machine ran, off which model and control, under what safety setup, and what it produced, both to prove the work and to make the next ROI call. The data the robot generates is only useful if it is captured against the job, so record it deliberately rather than letting it sit in a vendor portal.

Capture the robot and its OEM and software version, the model and control it ran from, the safety setup including the zone and the e-stop check and the operator, the productivity numbers, and the as-built output. Tie it all to the job and the date so a person reviewing it later can reconstruct what happened. A field platform such as FieldOS is a practical place to keep the safety, productivity, and as-built record together with the rest of the project documentation.

Common mistakes

• Expecting the robot to replace the trade instead of taking the repetitive work and freeing the crew for judgment.
• Feeding a bad model or the wrong control so the robot prints, drills, or grades the error perfectly.
• Putting a robot on a custom one-off or a congested space where the setup never pays back.
• Treating the robot as a fixed tool and ignoring that it is a moving hazard with a work zone and an e-stop.
• Running it with no trained operator and no time budgeted for setup, calibration, and maintenance.
• Expecting a return off a low-volume task instead of the repetitive high-volume work the machine is built for.

Field checklist

Standards and references

Construction robotics does not have one governing code the way the trades do, so the references split three ways: the machine, the model and control, and the safety. On the machine, the manufacturer or OEM governs. The capability, the accuracy, the operating envelope, and the maintenance all come from them, and their figures are claims to verify against your conditions, not guarantees. Hold every speed, accuracy, and ROI number to that standard before you build a plan on it.

On the model and control, the references are the same ones that govern the digital chain the robot builds from. The coordinated model comes out of the BIM and VDC process, and the survey control comes out of the layout and survey work, both covered in the sibling guides. The robot is only as good as those two, so the standards that govern them govern the robot's output by extension. A shared coordinate system between model and field is the practical link.

On safety, name the right standard for the machine. Autonomous and semi-autonomous earth-moving machinery falls under ISO 17757, remote-control machines under ISO 15817, and industrial mobile robots have their own safety standard, with OSHA's general requirements applying on top. Exoskeletons fall under emerging standards such as ASTM F48 and ISO 13482. The adopted standard, the manufacturer's instructions, and the site safety plan govern how the machine runs. The three lines to carry off this guide: robots augment the crew, they do not replace the trade; the robot is only as good as the model and the control it builds from; and pick the repetitive task, manage the moving hazard, and start with a pilot.

Terms and definitions

Construction robotics borrows terms from robotics, surveying, and the trades, so the same idea can read differently across a vendor sheet, a spec, and a safety plan. The definitions below are the ones used throughout this guide.

Construction robotics

The use of robots and automated machines to do dull, dirty, dangerous, and repetitive jobsite work, augmenting the crew rather than replacing the trade

Layout robot

A machine that prints or marks the building layout from the coordinated model directly onto the slab or deck, positioned against survey control

Autonomous / machine-control equipment

Earth-moving equipment guided to a design surface by GPS, either semi-autonomous with an operator in the cab or fully autonomous with none

Demolition robot

A remote-controlled tracked machine with a breaker, shear, or bucket that lets the operator work a dangerous or unstable structure from a safe distance

Exoskeleton

A wearable, usually passive frame that supports the back, shoulders, or arms to reduce strain on repetitive lifting or overhead work; it augments the worker, not the task

Model + control dependency

The rule that a robot is only as accurate and correct as the coordinated model and the survey control it builds from; a bad model or control yields bad output

Structured vs unstructured task

A structured task is repetitive and predictable enough for a robot to follow; an unstructured task is custom or unpredictable and needs a person

Human-in-the-loop

An arrangement where the robot runs the structured task and a person sets it up, supervises it, checks quality, and handles every exception

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is construction layout?
• What is BIM coordination?
• Do jobsite cameras stop theft?
• How do you prevent jobsite theft?
• Why do bridge decks deteriorate faster than the rest of the bridge?