Motionwell Automation provides robot integration in Singapore as a design-and-build service rather than a robot sale: the cell layout, the end-of-arm tooling, the part presentation, the guarding, the control system and the commissioning that turn a bought arm into a station that makes rate. Delivered work runs from a four-axis column palletizing platform placing cartons to plus or minus 1 mm at 6 to 10 cartons per minute, through two ABB IRB 6700-200/2.60 six-axis arms at 200 kg payload and 2.6 m reach on a battery module dismantling line, to seven JAKA collaborative arms purchased to date. Four Yamaha SCARA arms are being installed on a secondary packaging line in delivery through 2026. Grippers come from Schunk and DH Robotics with the application tooling designed in-house. The cells are designed, assembled and tested at our Woodlands Link facility, and the company has delivered more than 150 special purpose machines since 2014 under ISO 9001:2015 and bizSAFE Level 3.
Where we stand, said plainly before you read further: we do not manufacture robot arms, controllers, cameras or drives. We buy them. What we design and build is everything between the robot and your product, and the brand of the arm is usually the least interesting decision in the project. On most enquiries it has already been made, because your maintenance team stocks spares for one platform. Where a standard machine or a distributor’s packaged cell covers the job at a lower price, we say so during concept review rather than quote around it.
This page covers what a robotic system integrator actually delivers, how a robot brand gets chosen and who really decides it, which architecture suits which work, what a robot cannot fix, how a robot project runs stage by stage, and the safety obligations that arrive with a guarded cell. End-of-line stacking has its own selection logic on the automated palletizing systems page, robots tending instruments in a shared laboratory are covered under laboratory automation, and the safety workflow behind any guarded cell is on our machine safety and CE marking page. If you have a part drawing, a cycle time target and a layout, skip ahead and talk to an engineer.
What Does a Robot Integrator Actually Deliver Beyond the Arm?
An arm out of a crate does nothing useful. It does not know where the part is, what holds it, what happens when it is not there, or what to do when a person opens a door. Every one of those answers is engineering, and it sits outside the robot vendor’s scope.
| Scope element | Who supplies it | What it decides |
|---|---|---|
| Robot arm and controller | Bought: ABB, JAKA, Yamaha SCARA, Inovance, HitBot, Lanxin, or a customer-supplied arm | Payload, reach, speed envelope, safety modes available |
| End-of-arm tooling | Designed in-house; grippers from Schunk and DH Robotics, vacuum from SMC and Festo | Whether the robot holds every part, not most parts |
| Part presentation | Designed and built: feeders, trays, nests, escapements, conveyor squaring | Whether a pick is repeatable at all, and usually the largest single risk |
| Seventh axis or lift column | Designed and built: low-profile tracks, 100 mm profile height, plus or minus 0.05 mm repeat positioning | Whether one robot serves several stations instead of buying two |
| Guarding and safety circuit | Designed, built, validated and tested, including LVD and CE testing | Access strategy, stopping performance, residual risk |
| Cell controller and handshakes | Programmed on your plant standard: Allen-Bradley, Siemens, Mitsubishi, Omron, Beckhoff or Inovance | How the cell behaves when the line above or below stops |
| Vision and sensing | Specified and integrated: Keyence and Cognex cameras, SICK and ifm sensing | Whether the robot knows what it picked and where it put it |
The arm is one line on the bill of materials. The engineering is in the other rows, which is why two quotations naming the same robot can differ by a wide margin without either being wrong.
One row deserves a note. Buying a bigger robot to cover a long part usually costs more than putting a normal robot on a track, and it does nothing for the multi-station case. Our low-profile track exists because a conventional seventh axis eats 200 to 400 mm of vertical working height straight out of the arm’s usable envelope; the delivered configuration runs 2,850 mm of stroke with 1.5 tonnes of platform load and integrates with ABB IRC5 and OmniCore controllers as a coordinated external axis. Reach and cycle time are then verified in offline simulation before anything is built, because a pose that cannot be reached is a mechanical problem rather than a software one.
How Does a Robot Brand Get Chosen, and Who Really Decides?
Application requirements narrow the field. Payload, reach, speed, precision, safety mode and cost usually leave two or three platforms standing. After that the decision belongs to your plant rather than to us, and the deciding factor is normally the standard you already run: a platform your factory stocks spares for, trains on and has service cover for costs less to own than a cheaper unfamiliar brand, because the expensive parts of ownership are not on the purchase order. We ask which platform you run before we start selecting.
| Decision input | What it settles | Who owns it |
|---|---|---|
| Existing plant standard | Spares, training, programming environment, service response | Your maintenance and engineering teams |
| Payload at real centre of gravity | Whether the arm holds the part plus the tool at full extension | The integrator, from your part and tool data |
| Reach across the real layout | Furthest pick to furthest place, with fixtures and feeders in the way | The integrator, measured rather than estimated |
| Cycle time on the real part | Whether the station makes rate once vision, settling and handshakes are counted | The integrator, timed on your parts |
| Safety mode required | Fenced cell, or a collaborative application with a risk assessment behind it | Shared, decided by the ISO 12100 assessment |
| Customer-supplied arm | Whether the arm is bought at all | You, and it is a legitimate answer |
Our own position is multi-brand, and worth stating precisely because the phrase covers two different businesses. We buy and integrate ABB industrial and SCARA robots, JAKA collaborative arms with seven units purchased to date, Yamaha SCARA for lighter handling, and Inovance, HitBot and Lanxin platforms where the application suits them. Robot programming and end-of-arm tooling design for palletizing duty runs across FANUC, ABB, KUKA, Yaskawa, Yamaha and JAKA platforms. On mobile robots we have been the authorised SIASUN AGV distributor for Southeast Asia since 2017 and we integrate Youibot and MiR AMRs where a fixed route is the wrong answer, which is set out on the warehouse and intralogistics automation page.
One boundary is worth being exact about: we have integration experience on customer-supplied Universal Robots arms across three projects, where the customer had already standardised on that platform. That is a different thing from stocking them.
Which Robot Architecture Suits the Work You Are Automating?
Brand is the second question. Architecture is the first, and it is decided by the shape of the motion rather than by the industry the part belongs to.
| Architecture | Where it earns its place | Where it stops making sense |
|---|---|---|
| SCARA, four axes | Planar pick, place, insertion and dispensing at speed; stiff in Z where screwdriving loads it | Changing approach angles, or payload and reach outside roughly 1 to 20 kg and 200 to 1,000 mm |
| Six-axis articulated | Changing approach angles, confined access, force-controlled process work, heavy payloads | Simple vertical stacking, where the extra axes are cost you do not use |
| Four-axis column palletizer | Fixed stacking patterns at continuous duty; a purely vertical lift is mechanically efficient | Bag palletizing, multi-face labelling, or placing into constrained spaces |
| Collaborative arm | Shared space, occasional intervention, tight footprint, redeployment between lines | Cycle time under pressure, or when the hazard is the payload or the tool |
| AGV or AMR | Moving material between cells rather than working on it | Anything needing a tool at the end of an arm |
Three numbers make the trade concrete, and all three come from datasheets and market pricing rather than from a stopwatch on your parts. Published datasheet cycle times for a small SCARA run at roughly half those of a comparable six-axis arm on the same standard test move, which is why SCARA suits planar pick, place and insertion at speed. A four-axis column palletizer with a linear track typically costs 30 to 50 percent less than an equivalent six-axis solution for standard carton stacking, which is why we build our own column platform at 100 to 200 kg payload and 6 to 10 cartons per minute. And an ABB IRB 460 is specified where the line needs more than 12 cartons per minute, because it is rated above 2,190 cycles per hour. The stacking comparison sits in the robotic carton palletizing case study, and the planar case in our SCARA robot guide for electronics.
Treat any datasheet cycle time as a floor rather than a forecast. It is measured on a fixed gate-to-gate move with a dummy load and nothing else happening. Your real cycle adds vision acquisition and processing, settling before the vacuum releases, gripper actuation, the PLC handshake, and any wait on a conveyor or fixture. Those additions, not the arm, usually decide whether the station makes rate, which is why we time them on your parts before quoting a throughput figure.
Collaborative arms carry one constraint no purchase decision removes. An arm running as a power-and-force-limited application has its speed capped so any contact stays inside the ISO/TS 15066 limits, and that cap is the rate ceiling. You do not get around it by buying a faster arm. You get around it by adding guarding and speed, at which point you have built a small industrial cell with an expensive arm inside it.
What Can a Robot Not Fix?
This is the section most worth reading before you spend money. A robot repeats a motion. It does not create order that is not already there, and it does not absorb variation that arrives from upstream.
| What you see on the floor | What usually gets quoted | What is actually wrong | What fixes it |
|---|---|---|---|
| Parts arrive loose in a bin or a heap | A robot with a camera | Nothing presents the part in a repeatable pose | A feeder, tray or nest, or a flexible vision feeder that spreads parts flat before the pick |
| Runs well in commissioning, drops parts in month three | Retraining the robot | Floor markings wore, a rack leg was nudged, the tolerance stack moved | A docking feature, hand-eye vision measuring the real offset, and a compliant gripper for what is left |
| Robot waits, then rushes, then waits | A faster arm | Gaps in the upstream stream from rejects and stops | Accumulation and buffering sized as a design decision, not left over |
| Vision rejects climb across a shift | A higher resolution camera | Ambient light changed, or the fixture let the part sit differently | Controlled lighting, and a fixture that presents the part identically every cycle |
| Cycle time will not come down | A faster arm | The station waits on a cure, a test or a process dwell | Fix the process bottleneck; the arm is idle either way |
| The tool holds most parts | Tuning the vacuum | The tool was designed against the nominal part, not the worst one | Tooling designed against the full tolerance range, proven on samples |
Two engineering facts sit behind that table and both cost real money. First, feeding is chronically under-scoped: on our SCARA assembly lines a bowl feeder tuned to an awkward part can approach the price of the arm it feeds, and a feeder that presents a wrong-way part every fiftieth cycle costs far more than it looks like it should, because every one of those is either a jam or a reject. Second, inspection stations that fail in production usually fail because the ambient light changed or the fixture let the part sit differently, not because the camera was too small. How cameras get specified, lit and calibrated against that is on our machine vision inspection page.
The tolerance stack is the version that catches mobile robots, where an AMR parks to within about plus or minus 50 mm and an arm has to reach into a rack slot to a fraction of a millimetre. Skipping the vision correction is how a cell passes commissioning and starts dropping samples in month three, and the QA lab automation programme is built around exactly that layering.
What Does the End-of-Arm Tool Decide?
More than the robot does. The tool is where a cell is won or lost, and on a mixed line it is rarely something you can buy off a shelf unchanged.
Start with the payload budget, because this is where projects go wrong before anyone opens CAD. Rated payload is measured at the tool flange and your gripper hangs off that flange, so whatever the tool weighs comes out of the budget left for the part. Rated payload also assumes the load sits close to the flange, so a wide part pushes the centre of mass out and the resulting moment is what limits the arm rather than the mass alone. Do the arithmetic with your heaviest part and your real reach. If the heaviest item only just fits, it does not fit.
Tool families follow the part rather than the industry. Vacuum is the default on sealed cartons and flat rigid surfaces, with Festo pneumatic cylinders for side clamping where the top face is unreliable. Mechanical grippers from Schunk and DH Robotics cover trays, totes and bare product where vacuum has nothing to hold. Where the arm has to tighten something the tool becomes an instrument: on the SCARA sensor panel line electric screwdrivers work to a target torque set per fastener in the recipe, from 0.1 to 2.0 Nm, with final torque and rotation angle recorded against the panel serial number, and on the battery dismantling line torque-controlled nutrunners on the robot flange log torque and angle per fastener against the pack barcode.
Mixed-format lines get two honest options: one combination tool that compromises on every format, or a quick-change interface with a small tool rack, and changeover frequency decides which wins. We have delivered a cobot unloading station for a consumer goods distribution centre using a quick-change tool with a five-magnet electromagnet that switches between gripper and vacuum modes on its own, reconfiguring from a tool rack per carton on a barcode read rather than per shift.
One safety consequence belongs here rather than in the safety section, because it changes the mechanical design. End-effector geometry, workpiece edges and clamping forces all drive contact severity in a collaborative application, and long, thin or pointed tools concentrate force to the point where some geometries cannot pass a transient contact assessment at all.
How Does a Robot Integration Project Actually Run?
The sequence below is how we scope and run this work, and each phase has a review gate rather than a handover.
Concept design, 2 to 4 weeks. Process flow with cycle time per station, a preliminary 3D layout showing footprint and operator access, robot and major component selection, a risk register and a rough order of magnitude budget. This is the cheapest place in the project to change your mind, and reach gets checked here while layout is still free to move on a screen.
Detail engineering, 4 to 6 weeks. Full mechanical design in SolidWorks, electrical schematics, pneumatic layouts, control architecture, safety circuit design and the bill of materials. The largest schedule risk in the project appears at this gate: late finalisation of product design. If the part geometry changes after detail engineering begins, the tooling and usually the fixture change with it, so freeze the product first.
Fabrication and assembly, 6 to 8 weeks. Precision components machined to our own drawings by qualified machining partners, standard components sourced from approved suppliers, and the cell assembled and integrated in our own Singapore facility, following a documented build procedure with quality checkpoints.
Testing and commissioning, 2 to 4 weeks. Factory acceptance testing at Woodlands Link, witnessed by you, against the agreed cycle time, accuracy and repeatability. Then installation, site acceptance testing with real production material, operator training and the documentation pack. In regulated production, IQ and OQ protocols are prepared and PQ is supported with your quality team.
End to end, a simple one to three station cell runs 12 to 16 weeks from concept to factory acceptance, a medium multi-axis cell 16 to 24 weeks, and a complex integrated cell with vision and multiple robots 24 to 32 weeks, with 4 to 8 weeks added where GMP or cleanroom validation applies. The wider version of this process is in our guide to special purpose machine design.
What Safety Obligations Arrive With a Guarded Robot Cell?
They arrive with the robot, not with the fence, and they are the part of the scope most often discovered late.
Start from an ISO 12100 risk assessment and let the assessment drive the design rather than the reverse. ISO 10218-1 then covers the robot itself and ISO 10218-2 covers the robot system and its integration, which is the part a machine builder owns. Each safety function is rated for performance level under ISO 13849-1, and on both of the delivered robot cells described below that assessment landed on Performance Level d.
Two edition facts matter this year if you are writing a specification. ISO 10218-1:2025 was published in February 2025, the third edition and the first substantive revision since 2011; it adds robot classifications with matching functional safety requirements, safety-related cybersecurity requirements and end-effector guidance, and most of what ISO/TS 15066:2016 said about collaborative operation moved into Part 2. It references the 2023 edition of ISO 13849-1 for control system safety functions, so a design still documented against the 2015 edition will need its performance level calculations restated when the machine is reassessed.
The physical scope that follows is what we build and test: perimeter guard fencing, interlocked and solenoid-locking access doors, light curtains and laser scanners, safety-rated stop circuits, LVD and CE testing, and Ministry of Manpower lifting certification where the cell includes lifting equipment. On the delivered palletizing cell that means light curtains at the pallet removal openings, solenoid-locking door interlocks, and a safety circuit designed to ISO 13849 Performance Level d monitored by an ABB Pluto safety PLC, which watches the hardwired circuits independently of the robot controller. On the battery module dismantling line the controllers add ABB SafeMove2 functions on OmniCore for safe speed monitoring, safe standstill and safe axis range limiting, with an emergency stop that removes drive power within 10 ms.
Scanner and light curtain positions are calculated rather than chosen. ISO 13855 sets the minimum distance from the approach speed constant, taken as 2,000 mm/s up to 500 mm and 1,600 mm/s beyond, plus the stopping performance measured on the built cell. Guess it and you build a guard that looks right and stops too late.
For collaborative applications the distinction that matters is that there is no safe robot, only a safe application. A force-limited arm says nothing about the gripper, the workpiece or the fixture, and four collaboration modes are recognised, most cells using more than one in sequence. That assessment is worked through in our guide to collaborative robot safety standards.
Two Singapore points close this out. Singapore does not require CE marking, but the Workplace Safety and Health Act requires dangerous parts of machinery to be securely fenced and places duties on the occupier and on anyone supplying machinery for use at work; most multinationals here specify CE conformity as an internal standard anyway. And we are not a notified body. We do the machine builder’s share: the risk assessment, the architecture, the validation and the technical file.
How Does a Robot Cell Tie Into Controls You Already Run?
Much of the robot integration work we take on is a retrofit into a running plant rather than a greenfield line, so the interface is the project. The pattern we build to keeps motion and logic separate: on the delivered palletizing cell the robots run on a Mitsubishi iQ-R PLC architecture over EtherNet/IP, with the ABB IRC5 controller managing robot motion independently of the cell PLC. The cell PLC owns the handshakes with your line: part ready, cell ready, buffer full, fault, reject asserted. Agree who owns each of those signals in writing before either side is built, because an unowned handshake surfaces on your floor rather than in anyone’s factory.
Control system modernisation of existing machines, replacing ageing PLCs, servo drives and VFDs across Allen-Bradley, Siemens, Omron, Mitsubishi and Beckhoff platforms, is our largest line of work this year, so tying a new robot cell into old controls is routine here. Where an existing controller is too old to extend safely we say so and quote the replacement.
One regulatory point belongs in the kickoff meeting rather than in a change order. The European Commission states that the Machinery Directive applies to products placed on the market for the first time or when existing machinery is modified to such an extent that it becomes de facto new machinery. Adding a robot to a machine signed off years ago can cross that line, which puts the party doing it in the position of placing new machinery on the market with the conformity work that follows. Where the line sits is a judgement made per project rather than a number, and it has to be settled before the scope is fixed. The Directive applies to the EU market until 19 January 2027, after which Regulation (EU) 2023/1230 applies with no transitional period.
What Drives the Cost and Lead Time of a Robot Integration Project?
We do not publish prices for robot integration in Singapore, because the same nominal cell moves by a wide margin on decisions taken before hardware is ordered. What we can be specific about is which decisions move it.
| Cost driver | Why it moves the number |
|---|---|
| Part presentation | A part arriving in a nest is cheap; a part arriving loose brings a feeder, a tray system or a vision feeder |
| Part and format count | Each format may need its own tool, fixture and vision recipe, and format count costs more than it looks |
| Cycle time target | Sets the robot class, and decides whether guarding is optional or mandatory |
| Safety strategy | A fenced cell with scanners and locking doors versus a collaborative application with a clean risk assessment |
| Reach and multi-station duty | A seventh axis or lift column where one arm serves positions outside a single envelope |
| Vision scope | Presence checking alone versus guided picking, identifier reading and post-place verification |
| Retrofit archaeology | On legacy machines, working out what the existing circuits do before replacing them is frequently the largest single item in a retrofit safety scope |
| Validation and documentation | IQ/OQ/PQ evidence is a documentation project running alongside the build |
Lead time follows the phase table above: 12 to 16 weeks for a simple cell, 16 to 24 for a medium one, 24 to 32 for a complex integrated cell, plus 4 to 8 weeks where validation applies.
What Will We Not Take On?
Worth being direct, so nobody spends a month finding out.
We do not manufacture robot arms, controllers, cameras or drives, and outside our SIASUN AGV distributorship for Southeast Asia we are not a robot distributor. We do not issue CE certificates and we are not a notified body; we build to a specification and support your conformity work. We do not build production welding cells, and we do not sell software products, so the control and recipe software ships with the cell rather than as a licence. We are not a vision algorithm house either: we specify the camera, lens and light and prove the result on your parts.
And where a distributor’s packaged cobot cell or a standard machine serves you better and cheaper than a custom build, that is what we will say during concept review. The same position applies at the end of the line, where case packing and wrapping is a mix of what we build and what we integrate.
Which standard editions apply right now?
The editions below are the ones we design and document against on current projects. We check them on the date shown rather than assuming last year's edition still holds.
| Standard | Current edition | What it means for your machine |
|---|---|---|
| ISO 10218-1 — Robotics, safety requirements, Part 1: industrial robots | ISO 10218-1:2025 | Published February 2025, the third edition and the first substantive revision since 2011. It adds robot classifications with matching functional safety requirements, safety-related cybersecurity requirements, and end-effector guidance. Most of ISO/TS 15066:2016 on collaborative operation moved into Part 2. |
| ISO 13849-1 — Safety of machinery, safety-related parts of control systems | ISO 13849-1:2023 | The 2023 edition is the version referenced by ISO 10218-1:2025 for robot control system safety functions. Designs still documented against the 2015 edition will need their PL calculations restated when the machine is re-assessed. |
Frequently Asked Questions
What does a robot integrator deliver that the robot supplier does not?
Everything around the arm: the end-of-arm tool, the mechanism that presents the part the same way every cycle, the cell frame and layout, the guarding and safety circuit, the cell PLC and its handshakes with your line, the vision that confirms what the robot is holding, and the commissioning that proves cycle time on your parts. A robot vendor sells the arm, the controller and the programming environment. Motionwell buys those and designs the rest, with grippers from Schunk and DH Robotics and application tooling designed in-house. The useful test when comparing quotations: ask who owns the cycle time number on your actual parts.
Which robot brand should we specify for our cell?
Usually the one your maintenance team already stocks spares for and has been trained on. Selection is driven by payload, reach, speed, precision, safety mode and cost, but when two platforms both clear those requirements, the existing plant standard wins on cost of ownership because spares, training and support already exist. Motionwell integrates ABB, JAKA, Yamaha SCARA, Inovance, HitBot and Lanxin arms, deploys SIASUN AGVs and Youibot AMRs for mobile work, and has integration experience on customer-supplied Universal Robots across three projects. Tell us your site standard at enquiry stage so it lands in the concept rather than in a change request.
Will a robot fix a throughput problem?
Only if the constraint is the motion, and it usually is not. A robot repeats a motion accurately; it does not create order that is not already there. If parts arrive in a heap, the cost lands on the feeder or the tray rather than on the arm. If upstream stops leave gaps in the stream, the robot waits through every gap. If the station waits four seconds on a cure, a test or a dwell, a faster arm buys nothing at all. Count how many parts per minute reach the pick point in a usable presentation before you compare arm cycle times.