ROBOTIC.INDUSTRIES

Industrial Robotics

Delta Robots: Are 150 Picks per Minute Realistic?

Delta robots reach 150 to 200 picks per minute in a demo and 60 to 120 in production. Here is what removes the difference, and where a delta stops being the right machine.

Industrial robot wrist and gripper holding a machined metal part above a machine table
Industrial robot wrist and gripper holding a machined metal part above a machine table

Yes, but only on the manufacturer's test cycle. A parallel delta robot genuinely reaches 150 to 200 picks per minute on the standard 25 mm up, 305 mm across, 25 mm down adept cycle with a 100 g payload. In a real packaging line with vision, product variation and a moving belt, sustained rates land between 60 and 120 picks per minute, and the gap is almost entirely product presentation.

150 to 200picks per minute on the test cycle
60 to 120sustained rate in production
0.1 to 3 kgusable payload band
100 Gpeak accelerations quoted for light deltas

Why a delta is fast

The moving mass is tiny. Motors sit on the fixed base, the arms are carbon or aluminium links, and the only mass accelerated is the travelling plate plus the tool. A serial six-axis arm carries its own motors and gearboxes out along the chain, so it accelerates its own structure before it accelerates the part.

That structural choice produces the characteristic delta profile: extremely high acceleration over a short vertical stroke, a shallow working envelope, and a payload band of roughly 0.1 kg to 3 kg for high-speed models, extending to 8 kg for slower variants.

Where the demo rate goes

From test cycle to production rate
FactorTypical lossCause
Test cycle to real move distance10 to 25 %Real pick and place distances exceed 305 mm
Vision latency and tracking5 to 15 %Detection, tracking and handover to motion
Product not presented singulated10 to 30 %Overlaps and clusters are skipped
Gripper actuation10 to 20 %Vacuum build and release time
Belt speed mismatch5 to 15 %Robot waits or product escapes the window
Placement accuracy requirement5 to 20 %Settling before release into a tray pocket

Multiply those and a 180 per minute machine becomes a 90 per minute cell. None of it is the robot's fault, and none of it is fixed by buying a faster robot.

Vacuum release is the quiet cost. Building vacuum takes 15 ms to 40 ms and releasing it cleanly takes another 20 ms to 60 ms unless a blow-off pulse is used. At 120 picks per minute the whole cycle is 500 ms, so gripper timing is 10 % to 20 % of it. A larger ejector and a short positive blow-off is often the cheapest throughput improvement available.

Designing the line, not the robot

Delta cells are line problems. The decisions that set throughput are made upstream.

  • Singulation. Product arriving in clusters is product the robot cannot pick. A spreading conveyor, a vibratory feeder or a wider belt at lower speed usually buys more throughput than a second robot.
  • Belt speed. Too slow and the robot starves, too fast and product leaves the tracking window before it can be reached. The optimum is usually 60 % to 80 % of the speed at which the window closes.
  • Robot count. Multiple deltas over one belt share the flow, with each taking what the previous one left. Two robots rarely give twice the rate; 1.7 to 1.9 times is the realistic expectation.
  • Camera placement. The detection point must be far enough upstream that tracking is established before the first robot's window opens, typically 400 mm to 900 mm depending on belt speed.

Where a delta is the wrong machine

Delta against alternatives
RequirementDeltaBetter choice
Payload above 3 kg at high ratePoorSCARA or six-axis
Vertical stroke above 300 mmPoorSCARA or gantry
Tool must tiltOnly with a 4th axis, limitedSix-axis
Deep bin or box pickingPoor, envelope is a shallow domeSix-axis
Rate below 40 per minuteOverspecifiedSCARA
Washdown at 80 barAvailable in IP69K variantsDelta remains a good fit

The envelope is the usual disqualifier. A delta's working volume is a shallow dome, wide but not deep, so any task requiring reach into a container fails immediately regardless of speed.

Frequently asked questions

How many picks per minute can a delta robot really achieve?

150 to 200 on the standard test cycle with a 100 g payload, and 60 to 120 sustained in a real packaging line. The difference comes from product presentation, vision latency and gripper timing, not from the robot.

What payload can a delta handle?

High-speed models cover roughly 0.1 kg to 3 kg. Slower variants extend to about 8 kg, but the acceleration that makes a delta attractive falls away as payload rises.

Do two deltas double the rate?

No. Expect 1.7 to 1.9 times, because the second robot only picks what the first one left and both share the same tracking window. Improving singulation upstream often beats adding a robot.

Can a delta pick from a bin?

Not usefully. The working envelope is a shallow dome, so reaching into a deep container is outside its geometry. Bin picking is a six-axis task.

What limits throughput most in a delta cell?

Product singulation, followed by gripper actuation time. Clusters and overlaps are unpickable, and vacuum build plus release can consume 10 % to 20 % of a 500 ms cycle.

Sources

  1. ISO 9283, manipulating industrial robots, performance criteria and test methodsStandard cycle definition used for published pick rates
  2. ISO 8373, robotics vocabularyInternational Organization for Standardization, parallel kinematic machine definitions
  3. World Robotics 2025, industrial robotsInternational Federation of Robotics, food and packaging installation trends