Motionwell Automation integrates and builds several of the types of industrial robots set out below, and the classification that survives contact with a layout drawing is kinematic. The mix spans five families. Two six-axis arms at 200 kg payload and 2.6 m reach on a battery module dismantling line. SCARA arms holding ±0.01 mm repeat accuracy on vision-guided sensor panel assembly. Collaborative arms on assembly and laboratory cells. A linear gantry series built in three frame classes, from an aluminium extrusion frame through a welded steel single beam to a double beam bridge. And a four-axis column palletizing platform stacking cartons on a purely vertical lift column. Machines 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.
Two of the families below are our own products, the linear gantry series and the column palletizing platform. The arms are bought in, and what we design is what sits between the arm and your product: the tool, the presentation, the fixturing, the guarding and the sequence that ties them together.
This page is a classification. The collaborative versus industrial arm decision and the gantry versus six-axis choice for machine tending are argued elsewhere on the site. What follows is the layer underneath both of them, which is what each kinematic family sweeps in space and which layouts that shape suits.
How Many Types of Industrial Robots Are There, and What Separates Them?
The short answer. The dividing line is the kinematic chain, and everything commercial follows from it. A robot’s joints and links determine the shape of the volume its tool can reach, how payload behaves across that volume, whether the tool orientation is free or fixed, and how you buy more of it. Six families cover the industrial catalogue: six-axis articulated arms, four-axis SCARA arms, delta and other parallel machines, Cartesian gantry and truss frames, four-axis column palletizers, and collaborative arms. Application labels like “packaging robot” or “assembly robot” describe a job. Kinematics describes the machine, and it is the layer that decides whether a layout works.
Read the families as shapes on a drawing, and never as entries on a price list. Joints come in two kinds across all of them: a rotary joint, also called revolute, turns about an axis, and a linear joint, also called prismatic, slides along one. Every family below is an arrangement of those two, and the axis count is a tally of how many the machine has.
| Family | Axes | How the tool is carried | What the tool orientation does |
|---|---|---|---|
| Six-axis articulated | 6 | Serial chain of rotary joints, cantilevered from a base | Free within the envelope |
| SCARA | 4 | Two rotary joints in the horizontal plane, one vertical linear axis, one wrist rotation | Vertical approach only, with rotation about Z |
| Delta and parallel | 3 plus optional wrist axes | Actuators fixed to a frame above the work, linked to a common moving platform | Platform stays parallel to the base, with rotation about the vertical if fitted |
| Cartesian gantry and truss | 3 | Three linear axes at right angles on a fixed frame carried by columns | Fixed; the part arrives as the gripper picked it up |
| Column palletizer | 4 | Base rotation with a purely vertical lift column | Vertical approach with base rotation |
| Collaborative arm | 6 typically | Serial chain, as a six-axis arm | Free within the envelope |
Two things fall out of that table before any application is discussed. Serial chains, meaning everything except the delta and the gantry, carry every newton of payload as a moment through each joint in turn back to the base, which is why payload falls off with reach and falls off hardest where a long layout asks the arm to work. Frames, meaning the gantry and the delta, ground the load into structure instead, which changes what distance costs. That single distinction explains more purchasing mistakes than any spec sheet comparison.
Which Workspace Shape Does Each Type Sweep?
This is the part worth drawing on your own layout, because the envelope shape is what either fits your floor or does not.
A six-axis arm sweeps a shell around its base: a sphere with a hole in the middle where the arm cannot fold, and a pedestal in the way underneath. Reach is a radius, so it is bought in every direction at once including the directions you will never use. A SCARA sweeps a flatter version of the same thing, an annulus in one horizontal plane with a Z stroke through it. A delta hangs above the work and sweeps a shallow dome beneath its frame, small relative to the machine that carries it. A Cartesian gantry sweeps a rectangular prism drawn along a line, sized independently in X, Y and Z, which is the only envelope on this page you can draw straight onto a plan view without a compass. A column palletizer sweeps a vertical cylinder, a working radius swept through base rotation with a tall lift through it.
| Envelope shape | Family | Layout it suits | What breaks it |
|---|---|---|---|
| Shell around a base | Six-axis, collaborative | Stations arranged around one point, within one reach radius | Distance, because reach trades against payload |
| Horizontal annulus with a Z stroke | SCARA | Bench-scale stations where parts move from one flat surface to another | Any approach that is not vertical |
| Shallow dome above the work | Delta and parallel | A belt passing underneath, with the machine hung over it | Payload, and a workspace small for the frame that carries it |
| Rectangular prism along a line | Cartesian gantry and truss | Machines or stations in a straight row, loaded vertically | A scattered layout, or a part that must enter at an angle |
| Vertical cylinder | Column palletizer | Fixed stacking patterns around one position, or several on a track | Anything that is not a vertical lift and place |
The rule this whole page reduces to is a geometric one. Machines arranged in a row favour a beam, because a beam sells distance in modules and the payload at the far end is the payload at the near end. Machines arranged in a circle favour a wrist, because one base covers them all and the freedom of orientation is what the layout is asking for. Points scattered on both sides of an aisle suit neither, and the answer there is one machine per cell, or a rethink of the layout before a robot is quoted.
What Do the Published Class Bands Actually Compare?
Vendor datasheets give ranges per family, and read correctly they show relative strengths. None of them specifies a purchase. None of the figures below is a measurement from one machine.
| Factor | SCARA | Six-axis | Collaborative | Cartesian and gantry |
|---|---|---|---|---|
| Datasheet repeatability | ±0.01 to ±0.02 mm | ±0.02 to ±0.05 mm | ±0.03 to ±0.10 mm | ±0.01 to ±0.05 mm |
| Payload range | 1 to 20 kg | 3 to 300 kg and above | 3 to 25 kg | 5 to 100 kg and above |
| Reach | 200 to 1,000 mm | 500 to 3,000 mm and above | 500 to 1,300 mm | Frame-dependent |
| Relative speed on a planar pick | Fastest | Moderate | Slowest, because force limiting caps it | Moderate |
| Orientation freedom | Limited, four axes | Full, six axes | Full, six axes | Limited, three axes |
| Relative cost at equal payload | Lowest | Highest | High | Varies with travel |
Three cautions on reading it. Repeatability is not placement accuracy: it describes how tightly a machine returns to a taught point and says nothing about gripper slip, part tolerance, fixture wear or thermal drift, which usually dominate what you measure on the finished assembly. The cost row is ordinal, and it says nothing about the cell built around the arm. And a datasheet cycle time is a gate-to-gate move with a dummy load and nothing else happening, so vision acquisition, settling, gripper actuation and the PLC handshake all land on top of it, which is why we time those on real parts before quoting a rate.
Our own machines land where the frame families do not appear on vendor sheets at all. Across the linear gantry series the repeatability that can be held loosens step by step from the compact aluminium extrusion frame, to the welded steel single beam, to the double beam. The machines are not built to a worse standard as they grow; they get longer, heavier and more flexible, and structural deflection, thermal growth and drive stiffness all move the wrong way with size, so the figure is fixed per configuration once the span, the payload and the duty are known, and no figure is quoted for the family as a whole.
When Does a Six-Axis Articulated Arm Earn Its Extra Axes?
When the tool has to arrive at an angle, when the access is confined, when the process needs force along a contour, or when the payload is out of every other family’s band. Those four cases are what you are paying the extra joints for, and outside them the axes are cost you do not use.
Angle and access are the common pair. A chuck that sits tilted, a part that flips between the first and second operation, a fastener on a face that is not the top face: three linear axes have no answer to any of them, and no amount of travel substitutes for a wrist. Confined access is the same argument in a smaller space, where the arm has to fold around an obstruction to reach the target.
Process force along a path is where the family is alone. Our force-controlled robotic grinding of CFRP composite panels holds 100 N of constant contact force, developed with A*STAR SIMTech, and holding a tool normal to a curved surface while maintaining that force is six-axis work by definition. Grinding, deburring and dispensing along a contour all sit here.
Payload is the fourth case and it is simply arithmetic. On the battery module dismantling line the modules weigh 30 to 80 kg, the pack carries 400 V and above, and two six-axis arms at 200 kg payload and 2.6 m reach do the fastener removal, busbar disconnection and module extraction inside fenced stations. The arms are sized well above the heaviest module, because the tool that has to hold a live module without crushing it weighs more than the grippers a catalogue payload figure assumes. Nothing else in the classification reaches that band. At the fast end of the same band the four-axis palletizing arms take over, and they are worth reading as a trade. Dropping the wrist buys a stiffer structure and a shorter cycle at heavy payload, and it costs you every pattern that needs a case turned about a horizontal axis, for the life of the cell. That is where a line whose sustained rate outruns a lift-and-place column goes, and the thing to settle before it is bought is whether the stacking pattern is genuinely fixed or merely fixed today.
One number no datasheet states plainly. Rated payload is measured at the tool flange and assumes the load sits close to it, so the gripper comes out of the budget before the part gets any, and at full reach the moment is what limits the arm. Read the payload-versus-centre-of-gravity chart before you choose a model, and the end-of-arm tooling design page for what that costs in practice.
Where Does a SCARA Beat a Six-Axis Arm on the Same Bench?
On horizontal-plane work at speed, and the mechanical reason is a split of duties. The two rotary joints carry the horizontal load while the vertical axis only handles gravity and insertion force, so the arm can fly across a tray and still press a connector home without deflecting. 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.
Our delivered work in the family is the vision-guided sensor panel assembly cell built on a SCARA arm at ±0.01 mm repeat accuracy, where components are picked, aligned under vision feedback and inserted into panel assemblies, written up in the SCARA panel assembly case study. Packaging work sizes the same family from the opposite end: reach and speed follow carton and case geometry, with placement accuracy barely entering the sizing, and where several arms share one line they get sequenced against the station that limits the line, never tuned individually to their own best cycle, because an arm finishing early only builds a queue in front of the next one.
The constraint is the wrist. It rotates about Z and nothing else, so the approach vector is always vertical, and every argument for and against the family comes back to that. Tilted insertions, contoured surfaces and multi-face work belong on a six-axis arm. Long straight travel belongs on a frame. Chip-scale components at SMT rates belong on a placement machine, which will beat any general-purpose robot at that job. The full sizing method is in our SCARA robot guide for electronics manufacturing.
What Do Delta and Parallel Robots Do That a Serial Arm Cannot?
They move a very light tool very fast, and the reason is structural and not a matter of better motors.
In a serial arm each motor carries the mass of every joint downstream of it, so the machine accelerates its own arm before it accelerates the part. A delta inverts that. The actuators are fixed to a frame above the work and drive three linked arms that meet at a common moving platform, so what accelerates is the platform, the links and the tool, and the motors stay still. Stiffness comes from the closed loop of links instead of from a cantilever, which is a different and generally more efficient way to buy it.
Two consequences follow directly. The payload band is light, because a structure optimised for low moving mass is not a structure optimised to carry load. And the workspace is a shallow dome beneath the frame, small in relation to the machine that carries it, so the family belongs above a moving belt and never around a set of stations. The moving platform stays parallel to the base, which makes the orientation behaviour closer to a SCARA than to a six-axis arm: vertical approach, with rotation about the vertical axis only if a wrist axis is fitted.
Where that lands commercially is high-rate picking from a belt where the parts are light and the placement is planar: small moulded parts, sachets, confectionery, and anything arriving faster than a serial arm can index against. A delta is normally bought as a packaged cell with its vision and its belt tracking already matched to the frame, so the sizing figures come from the machine’s own maker and the integration work is the belt either side of it.
When Is a Cartesian Gantry the Right Frame Instead of an Arm?
When the work is in a line, when the distance is long, and when the part can enter the way it was picked up.
Distance is what a beam sells cheaply. The X axis runs on rack and pinion in sections that join without a fixed limit, so two more machines on the row means more beam and rail instead of another robot, another controller, another tooling set and another guarded footprint. Payload does not fall off along the travel the way reach does on an arm, because the load goes into the beam, the beam into the columns and the columns into the floor. Our own series is built in three frame classes on that logic: a compact aluminium extrusion frame for light parts, a welded steel single beam for heavier loads down a long row, and a double beam bridge with its own walkway and handrails where the span and the load are largest. One steel single beam with dual Z heads services a row of CNC machines, one head placing a blank while the other clears the finished part.
What sizes the beam is not the motor. Deflection between supports scales with the cube of the span and inversely with the second moment of area of the section, so doubling an unsupported span is roughly a factor of eight in droop at the same section, and the recovery comes from section depth or another column. The moving half costs the cycle: the beam is a spring and the carriage is a mass, so every stop excites an oscillation and the Z column cannot descend into a fixture until it decays. A quotation that answers a longer span with a larger drive is answering the wrong question. The structural detail behind all of this sits on the linear gantry and truss robot page.
What you give up is the one thing three linear axes cannot buy, which is attitude. The part arrives in the orientation the gripper picked it up in, which pushes cost upstream into trays, magazines and locating fixtures because the pick has to be right the first time. Where one station in a row needs rotation, a passive flip fixture or a single arm at that station is usually cheaper than replacing the beam.
Why Does a Palletizer Only Need Four Axes?
Because the motion it performs is a vertical lift, a rotation and a place, and buying orientation freedom for that is buying an axis you never command.
Our column palletizing platform is four axes: a base rotation covering the infeed and the pallet stations on either side, a purely vertical lift column doing the height, a horizontal arm setting the reach, and a wrist rotation setting the case angle in the pattern. A dedicated four-axis palletizing kinematic costs meaningfully less than a six-axis arm doing the same fixed stacking pattern, and the saving is structural, not a discount: fewer reducers, smaller servos because the counterweighted column lifts instead of a cantilever holding the load out, and fewer controlled axes through the controller and the safety chain. The placement tolerance on that machine was set by the case and never by accuracy for its own sake, because corrugated cases hold ±2 to 3 mm on their own outside dimensions, and that is the kind of reasoning that should decide every accuracy specification on a machine.
Where the four axes stop is equally clear: bag palletizing, multi-face labelling and placing into constrained spaces all need something the family does not have. Where the line rate outruns what a lift-and-place column sustains, the answer is a faster palletizing arm and never a different kinematic. And where the pattern varies enough that the fixed geometry stops being an advantage, the six-axis arm is back in the comparison. The selection logic for end of line sits on the automated palletizing systems page and the delivered cell in the robotic carton palletizing case study.
Is a Collaborative Arm a Separate Type at All?
Kinematically, no. A collaborative arm is a six-axis serial chain with the same envelope shape as any other six-axis arm, and it belongs in this classification as a safety architecture. What it adds is safety-rated functions intended for working close to people as standard, where an industrial arm is specified for a guarded cell unless safety-rated speed, standstill and axis limiting are configured on its controller.
The consequence that matters when you are picking a type is that the force limit is a rate ceiling, not a feature you can tune away. An arm running as a power-and-force-limited application has its speed capped so any contact stays inside the contact limits, and you do not get around that by buying a faster collaborative arm. Where floor area is the binding constraint we mount the collaborative arms overhead, which frees the floor for conveyors and operator access while keeping full reach, at the cost of easy maintenance access.
The standards apply across the whole classification. Start from an ISO 12100 risk assessment; ISO 10218-1 then covers the robot itself and ISO 10218-2 the robot system and its integration, with each safety function carrying a required performance level under ISO 13849-1, calculated and validated. The 2025 third edition of ISO 10218-1, published in February 2025 and the first substantive revision since 2011, added 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. If you are writing a specification now, check which edition your contract names. The rest of that decision, including where a cobot is the wrong purchase, is on the cobot versus industrial robot comparison.
How Do You Read a Layout and Choose the Type?
Five questions, in this order, answered on a drawing.
- Where do the work points sit? Mark every pick and every place. A line, a circle, a belt or a scatter is the first fork, and it narrows the field further than any datasheet does.
- Does the part change attitude between the pick and the place? If it does, either buy the wrist or design the flip, and price both. If it does not, the frame families are open to you.
- What is the true payload? Heaviest part plus tooling and cabling, checked against the payload curve at your real centre of gravity and reach. If it only just fits, it does not fit.
- What sustained rate does the line have to hold? Sustained, and never the datasheet cycle. If the rate needs full speed with nobody inside, the safety architecture is decided too.
- How often does the layout itself change? A structure is sized to the row it was drawn for. An arm on a pedestal relocates and is re-taught.
| What the layout looks like | Start from | Why |
|---|---|---|
| Stations around one point, mixed approach angles | Six-axis articulated arm | Free orientation inside one reach radius is exactly what the layout asks for |
| A straight row of machines, parts loaded vertically | Cartesian gantry or truss | Distance is modular on a beam and the floor underneath stays clear |
| Small parts moving between two flat surfaces at speed | SCARA | The horizontal joints carry the load and the Z axis stays stiff |
| A belt of light parts at high rate, planar placement | Delta or parallel machine | Low moving mass is a structural advantage no serial arm matches |
| Fixed stacking patterns at continuous duty | Four-axis column palletizer | The unused axes on a six-axis arm are cost without a job |
| A person genuinely inside the space through the shift | Collaborative arm | The only mode that survives continuous human presence, at a capped rate |
| Several stations outside one reach, orientation still needed | Six-axis arm on a linear track | Distance without giving up the wrist |
That last row is the option a two-way comparison tends to leave out. Buying a bigger arm to cover distance usually costs more than putting a normal arm on a track, and it does nothing for the multi-station case. Our low-profile design runs 100 mm of overall track height at ±0.5 mm repeat positioning with no floor pit, because a conventional seventh axis eats 200 to 400 mm of vertical working height before the arm has done anything, and the delivered configuration carries 1.5 tonnes of platform load over 1.5 m or more of stroke on a 3 m track as a coordinated external axis. The detail is on the robot seventh axis linear track page.
What Settles the Choice in the End?
Choosing among the types of industrial robots is a geometry exercise before it is a purchasing one. Each family sweeps a characteristic volume, and that volume either matches the shape of your work or it does not: a shell around a base, a horizontal annulus, a dome above a belt, a prism along a line, a vertical cylinder. Machines arranged in a row favour a beam. Machines arranged in a circle favour a wrist. Everything else on a datasheet is a refinement inside that first decision.
What the type does not decide is whether the cell works. A robot repeats a motion; it does not create order that is not already there. If parts arrive loose, the cost lands on the feeder or the tray. If upstream stops leave gaps in the stream, the robot waits through every gap. If the station waits four seconds on a cure or a test, a faster machine of any family buys nothing at all. Count how many parts per minute reach the pick point in a usable presentation before you compare cycle times, and read the scope around the arm on our robot integration services page.