Applications: Palletizing, Welding and Machine Tending

Pallet patterns, welding travel speed and heat input, machine tending cycles, dispensing and painting, with computed examples.

Choosing and sizing an application

ApplicationTypical robotWhat decides success
Palletizing and depalletizingFour- or six-axis, 20 to 300 kg; cobot palletizers for light loadsPattern, gripper, speed of the in-feed, pallet and stack stability
Arc weldingSix-axis with a torch and a positionerFit-up of the parts, torch access, seam tracking, fume extraction
Machine tendingSix-axis or cobot beside a machine tool, or on a trackDoor and chuck interface, the machine’s cycle, double gripper
Dispensing and sealingSix-axis or cartesian with a metering pumpA constant bead along the path, including corners
Painting and coatingSix- or seven-axis, explosion-proof for solventsTransfer efficiency, path and overlap, booth conditions
Assembly and screwdrivingSCARA or small six-axis, cobotPositioning accuracy, compliance, feeding
Deburring, grinding and polishingSix-axis with force controlConstant contact force and tool wear compensation
Packaging and sortingDelta or SCARA with visionPick rate, conveyor tracking, hygiene
Inspection and measurementSix-axis carrying a camera or probeAccuracy and repeatability of the path (see Vision, Calibration and Accuracy)

Each of these has the same chain of decisions: the robot type (see Robot Types, Specifications and Selection), the tool (see End-of-Arm Tooling and Grippers), the cell (see Cell Design, PLC Integration and Feeding) and the safeguarding (see Safety: Standards, Safeguarding and Cobots). This page adds what is special to each application, with worked numbers.

Palletizing

Palletizing stacks boxes or bags on a pallet in a pattern. The robot picks from a conveyor (one or more boxes at a time) and places on the pallet according to a pattern that is built from the box size and the pallet size. The pattern decides the stability of the load and how many boxes a pallet holds.

Layer patterns on a palletLong side along the pallet: 6 boxesarea used 68 %Best pattern found: 8 boxesarea used 90 %Boxes 450 x 300 mm, pallet 1 200 x 1 000 mm. Blue: long side along the length; yellow: turned 90 deg. Real layers also alternate for interlock.
Layer patterns for 450 × 300 mm boxes on a 1 200 × 1 000 mm pallet: all boxes in one orientation, and the best arrangement found by turning some of them.

Worked example: boxes of 450 × 300 × 250 mm, 8 kg each

  • Boxes per layer. With the long side along the pallet length, 2 fit along 1 200 mm and 3 across 1 000 mm: 6 boxes, using 68 % of the pallet area. Turning the boxes by 90° gives 8 (90 % of the area), the best simple pattern found for this size.
  • Layers. With a pallet height of 150 mm and a maximum stack height of 1 800 mm: ⌊(1 800 − 150) ÷ 250⌋ = 6 layers, for a stack 1 650 mm high.
  • Per pallet: 8 × 6 = 48 boxes, 384 kg of boxes, inside the 1 000 kg limit.
  • Interlocking. Alternate the layer pattern (turn the pattern by 180° or mirror it) so that vertical joints do not line up; a column of boxes with no overlap is unstable.
  • First four box positions (center, in pallet coordinates): the program calculates each place position from the pattern, not from taught points.
BoxCenter x, y (mm)Turn
1150, 22590°
2150, 67590°
3450, 22590°
4450, 67590°
  • Rate. A robot that picks 2 boxes at a time with a 5.0 s cycle moves 24 boxes a minute, so it fills a pallet in 2.0 min. Changing a full pallet takes 45 s, which cuts the effective rate to 17.5 a minute (27 % lost). Use two pallet stations so that the robot works on one while the other is changed.
  • Stability. The stack’s center of gravity is about 825 mm high; the stack tips when it is inclined by atan((1 000 ÷ 2) ÷ 825) = 31°, a figure that falls quickly when the layers are loose or the pallet is accelerated, so wrap or strap loads that will travel.
  • Gripper: vacuum with a clamp for boxes, a clamp for bags, a fork for sacks; check the load chart (see Payload, Inertia, Speed and Cycle Time); the layer gripper picks whole layers.
  • Pallet and slip sheets vary: sense the pallet position and height, and leave a margin for the pallet’s tolerance.
  • Safety: the pallet station needs a safe way for a person to remove a full pallet without entering the robot’s area: a light curtain with muting or a rotating station (see Safety: Standards, Safeguarding and Cobots).
  • Cobot palletizers come with a pattern wizard and are limited by payload and speed; check the cycle against the line.

Arc welding

A welding robot carries a torch along a seam at a controlled travel speed while the power source supplies current and wire. The seam quality depends on the heat input, the wire deposition and the fit-up of the parts.

Heat input Q = η U I ÷ v (volts × amperes ÷ mm/s gives J/mm)

Heat input falls as the robot travels faster0246810120.00.51.01.52.0travel speed (mm/s)heat input (kJ/mm)example upper limit 1.0 kJ/mmminimum speed 3.8 mm/s
Heat input against travel speed for 24 V, 200 A and an arc efficiency of 0.8.

Worked example: a 5 mm fillet weld

  • Wire deposited. A 1.2 mm wire fed at 5 m/min gives a volume of π ÷ 4 × 1.2² × 5 000 = 5 655 mm³/min, or 44 g/min of steel (density 7.85 g/cm³): about 2.7 kg per hour.
  • Travel speed. A 5 mm leg fillet has a cross-section of ½ × 5² = 12.5 mm². At 95 % deposition efficiency the travel speed that fills it is 5 655 × 0.95 ÷ 12.5 ÷ 60 = 7.2 mm/s.
  • Heat input. At 24 V and 200 A with an efficiency of 0.8: Q = 0.8 × 24 × 200 ÷ 7.2 = 536 J/mm (0.54 kJ/mm).
  • If a material standard sets an upper limit of 1.0 kJ/mm, the robot may not travel slower than 3 840 ÷ 1 000 = 3.8 mm/s. Traveling twice as fast (14.3 mm/s) halves the heat input to 268 J/mm but leaves the fillet underfilled unless the wire speed doubles.
  • Arc-on time. A 600 mm seam takes 84 s of welding; with 22 s for loading, positioning and moves, the arc is on for 79 % of the cycle. Raising arc-on time (a positioner, two stations, fast moves) is the main way to raise productivity.
  • Fit-up and tolerance. A robot follows the programmed path: a gap or a mismatch of 1 mm that a welder would adjust for needs touch sensing, laser seam tracking or through-the-arc tracking (see Vision, Calibration and Accuracy).
  • Torch access and orientation: work angle and travel angle control the penetration; check access and collisions in simulation; use a collision sensor on the torch (see End-of-Arm Tooling and Grippers).
  • Positioners (one- or two-axis tables) put the seam in the best, downhand position, and can move in coordination with the robot.
  • Torch maintenance: a tip, nozzle and liner changer and a cleaning station, with a tool center point check after each change (see Coordinate Frames and Kinematics).
  • Safety: arc flash and fume: screens and fume extraction, protection for people near the cell; the robot is welding behind a guard with a door interlock (see Safety: Standards, Safeguarding and Cobots).
  • Weld procedures and qualification (WPS, ISO 15614 or AWS D1.1) apply to robotic welds just as to manual ones.

Machine tending

A robot loads parts into a CNC machine, a press or a molding machine and takes the finished parts out. The aim is to keep the machine cutting: the robot does the unproductive loading time. The robot must be able to open the door, reach the chuck or fixture, load and unload without crashing, and signal the machine (see CNC Machining for the machine side).

Worked example: a lathe with a 90 s cycle

  • With a single gripper, the robot unloads the finished part, then loads a new one: the machine stands still for 12 s each cycle. Machine utilization is 90 ÷ (90 + 12) = 88.2 %, 35.3 parts an hour.
  • With a double gripper (the robot takes the finished part out with one gripper, swaps and puts the new blank in with the other) the machine stops for only 7 s: utilization 92.8 %, 37.1 parts an hour, which is 5.2 % more from the same machine.
  • How many machines can one robot serve? If each machine needs 12 s of service and the robot needs 4 s to move between machines, the robot’s cycle per machine is 16 s, and in principle it can look after (90 + 12) ÷ 16 = 6.4, that is up to 6 machines. In practice, stagger the machines, allow for stoppages, and serve about 3 (the robot is then busy 47 % of the time).
  • Interface with the machine: an automatic door, a clamping signal from the chuck, a program-start and an end-of-cycle signal, and a safe zone shared between the robot and the machine (see Cell Design, PLC Integration and Feeding).
  • Parts in, parts out: trays or drawers for blanks, conveyors or trays for finished parts, a gauge or an air blast to clean the chuck, and a check that a part is gripped and seated.
  • Collisions: program the robot inside the machine slowly the first time, and use the machine’s own safe-zone interlocks.
  • Cobots suit low-volume tending: they can be wheeled to a machine and taught in minutes, at a lower speed.

Dispensing, sealing and gluing

A robot carries a nozzle along a path while a metering pump supplies adhesive or sealant. The bead volume per millimeter is the flow divided by the speed, so the flow must follow the speed, including at corners and starts.

Worked example: a bead of 3 mm²

  • At 200 mm/s a 3 mm² cross-section needs a flow of 3 × 200 = 600 mm³/s (0.6 mL/s).
  • A 1 200 mm path takes 6.0 s and uses 3.6 mL per part, so a 310 mL cartridge lasts about 86 parts.
  • The corner problem: if the pump keeps a constant flow while the robot slows to 50 mm/s for a corner, the bead cross-section grows to 600 ÷ 50 = 12 mm², four times thicker. Use a pump that follows the robot’s speed (the robot sends a speed-proportional signal), and start the flow a moment before the move.
  • Keep the nozzle height and angle constant: the bead shape depends on them, so the tool center point and the path accuracy matter (see Coordinate Frames and Kinematics).
  • Check the bead with a vision system (width, position, gaps) where the quality matters.
  • Material handling: temperature, pot life, and purging; the nozzle needs a cleaning station.

Painting and coating

Worked example: paint used on a part

A dry film of 60 µm over 1.5 m² needs 60 × 10−6 × 1.5 = 90 cm³ of solid paint on the surface. With a transfer efficiency of 65 % (the fraction of sprayed paint that lands on the part), you must spray 138 mL for each 90 mL applied: the rest is overspray. Raising the efficiency from 65 % to 85 % (electrostatic or rotary atomisers) cuts the paint used by 24 %.

  • Explosive atmosphere: solvent paints need an explosion-proof robot (ATEX or the Class I Div 1/Zone 1 equivalents), purged and certified. A standard robot must not be used.
  • A smooth path with constant speed and distance gives an even film; blends and joint moves are used with care.
  • Booth airflow, cleanliness and curing are part of the cell design.

Assembly and screwdriving

Worked example: a cover with 8 screws

Each screw needs 0.5 s to move to the position, 0.6 s to feed and pick up the screw (a screw feeder or a suction bit) and 1.2 s to drive it: 2.3 s. For 8 screws that is 18.4 s of cycle. Driving the screws with two spindles at once, or feeding by blowing screws through a tube to the nozzle, cuts the time with no faster robot.

  • A SCARA is the classic assembly robot: fast, rigid in the plane and compliant in the horizontal plane (selective compliance), which is what inserting a part needs.
  • Compliance and force sensing solve the insertion problem (see End-of-Arm Tooling and Grippers and Vision, Calibration and Accuracy).
  • Monitor the screwdriving (torque against angle) and record it for quality; use the controller’s result to stop the line on a fault.

Deburring, grinding and polishing

Finishing needs a constant contact force, not just a position, because the surface and the tool wear. The usual solution is a force-controlled spindle or a compliant end effector with a pneumatic force regulator, or a force-torque sensor in the loop (see Vision, Calibration and Accuracy).

Worked example: contact pressure

A force of 20 N over a contact patch of 40 mm² gives a pressure of 0.5 N/mm². If the tool wears and the patch doubles, the pressure at the same force halves: the process changes without any change in the program. Control the force, and measure the tool’s wear (a counter, a sensor or a regular check) rather than the number of parts.

  • Dust, noise and sparks need an enclosure and extraction; check the fire risk.
  • Offline programming from CAD is common here, which needs a calibrated robot and tool (see Vision, Calibration and Accuracy).

Packaging, inspection and others

  • Packaging and sorting use deltas and SCARAs on conveyors with vision (see Vision, Calibration and Accuracy on tracking); hygienic designs for food and pharmaceuticals.
  • Inspection: a robot carrying a camera or a probe measures large parts (a car body, a casting). The robot’s accuracy sets the measurement accuracy, so calibrate it or use a measuring arm.
  • Laser and waterjet cutting, laser welding: need accurate paths and special safeguarding for the beam (see Laser Cutting).
  • Injection molding take-out: a fast, simple cartesian robot on top of the press (see 3D Printing for the additive side).
  • Logistics: mobile manipulators and truck unloading, with 3D perception and learned grasping (see Machine Learning).
  • Lab automation and pharma use small precise robots in cleanrooms.

Related

FAQ

How do I work out how many boxes fit on a pallet?

Divide the pallet length and width by the box dimensions in each orientation, round down, and try the patterns with some boxes turned; then multiply by the number of layers allowed by the height limit. Check the weight limit.

Why does my welding robot make poor welds when the manual weld was fine?

A robot follows the programmed path and cannot see a gap or a mismatch. Check the fit-up, add seam tracking or touch sensing, and check the tool center point and torch angle.

Is a double gripper worth it on a machine tending cell?

Usually, if the machine’s cycle is short: it halves or better the time the machine stands idle for loading, which raises output from the same machine at a small cost for the tool.

Why does my sealant bead change at corners?

The pump keeps a constant flow while the robot slows down, so the bead gets thicker. Use a pump that follows the robot’s speed and start the flow before the move.

References

  • AWS D1.1 / D1.2 and ISO 15614 (welding procedure qualification); the welding power-source makers’ guides to robotic welding.
  • EUR-pallet and ISO pallet standards (ISO 6780); the palletizing and packaging machinery makers’ pattern guides; ISO 10218-2 for the cell safety.
  • Groover, M. P., Automation, Production Systems and Computer-Integrated Manufacturing, Pearson; Nof, S. Y. (ed.), Handbook of Industrial Robotics, Wiley.

Last reviewed: October 2026. Rates, speeds and efficiencies are typical and illustrative: use your own data and the equipment makers’ figures.