There is a robot for almost every task, which is helpful right up until you have to choose.
The useful way to think about it is not "which robot is best" but "which shape of motion does my process need". Each type trades speed against flexibility against payload against footprint. Pick the one whose trade-offs match your problem and the rest of the design gets easier.
SCARA
Selective Compliance Assembly Robot Arm. Four axes. Rigid vertically, compliant horizontally.
Built for fast, repeatable work on a plane: pick and place, small parts assembly, screwdriving, dispensing. A SCARA moving parts between two points on a table will beat a six-axis arm doing the same job, and cost less.
The limitation is the same as the strength. It cannot approach from an angle. If your part needs to be tilted, rotated, or presented at anything other than straight down, a SCARA is the wrong tool.
Choose it when: high speed, high repeatability, everything happens on one plane.
Delta
Three or four axes, mounted overhead, with the motors in the base and lightweight arms reaching down. That geometry is what makes it quick, because there is very little mass moving.
Delta robots are the fastest thing in the category for light payloads. High-speed picking off a moving conveyor, usually with vision, usually food, pharmaceutical, or small consumer goods.
Payload is low, often a kilogram or two, and the working envelope is a shallow dome rather than a sphere.
Choose it when: very high rate, very light parts, picking from a conveyor.
Six-axis articulated
The one most people picture. Six degrees of freedom, which means it can reach a position from almost any angle.
That flexibility is why it dominates. Machine tending, welding, material handling, assembly, dispensing, palletising. It reaches around obstructions, approaches at compound angles, and handles payloads from a few kilograms to several tonnes.
The trade-offs are real. More axes means more complex programming and generally slower cycles than a SCARA or delta doing simple planar work. You pay in speed for motion you may not need.
Choose it when: the process needs orientation, not just position. Which, on most custom automation, it does.
Heavy payload
Still six-axis, but a different engineering conversation. Machines rated from several hundred kilograms into the thousands.
On one aerospace cell we used a FANUC M-2000iA/2300, rated at 2,300 kg, to handle landing gear and structural components.
The robot is the straightforward part. The floor is not. The mass and dynamic loading of a machine that size, accelerating and decelerating with a full part in its grip through thousands of cycles, imposes structural demands well beyond a normal industrial slab. On that project the pit was excavated to six feet and reinforced throughout with rebar.
Choose it when: the part is genuinely heavy. Then budget for the foundation, not just the robot.
Collaborative robots
Designed with force and speed limits so they can work near people without full guarding.
Genuinely useful for low-rate tasks where fencing is impractical, where the robot works alongside an operator, or where the cell has to move around. The catch is that the safety case is not automatic. "Collaborative" describes the robot, not your application. A cobot holding a sharp tool or a heavy part still needs a risk assessment, and it may still need guarding.
They are also slower. Safety limits are speed limits.
Choose it when: low rate, shared space with people, and the risk assessment supports it.
Cartesian and gantry
Linear axes at right angles. Not glamorous, and often correct.
Large working envelopes, high payloads, simple motion, and often cheaper per cubic metre of reach than an articulated arm plus a track. Good for palletising, machine loading over a long span, and anything where you need to cover a big rectangle.
Choose it when: the envelope is large and rectangular, and the motion is simple.
Purpose-built mechanisms
Sometimes the answer is not a robot.
A dial table indexing parts through six stations will outperform a robot doing the same sequence, more cheaply and in less floor space. A well-designed escapement, a pick-and-place unit, or a simple cam mechanism can beat a programmable arm on a fixed, high-rate task.
Robots earn their place through flexibility. If the task will never change and the rate is high, flexibility is something you are paying for and not using.
See In praise of an automation workhorse: the dial table.
A shortcut
| If the deciding factor is | Start with |
|---|---|
| Raw speed, light parts, off a conveyor | Delta |
| Speed on a single plane | SCARA |
| Orientation and reach around obstructions | Six-axis articulated |
| A genuinely heavy part | Heavy-payload six-axis, and a serious foundation |
| Working beside people at low rate | Collaborative, with a real risk assessment |
| A large rectangular envelope | Cartesian or gantry |
| Fixed sequence, very high rate, never changes | Purpose-built mechanism, possibly no robot |
The thing that actually decides it
In practice, robot selection is rarely the hard part of a project. Tooling is.
The robot moves the part. The end-of-arm tooling has to grip it, hold it through the process, and cope with it arriving at the edge of its tolerance. On most cells we build, the tooling takes more engineering than the robot choice, and it is where the project is won or lost.
Pick a robot that comfortably covers your payload and reach with margin, and spend the argument on the gripper.