Industrial Robotics
How to Calculate Robot Cycle Time Before You Sign
A method for estimating robot cycle time from a datasheet, with the acceleration and settling terms that vendor demos leave out, plus a worked machine tending example.
Estimate each move as acceleration, constant velocity, deceleration and settling, then add process time that no robot can shorten. On short moves the arm never reaches its rated speed at all, which is why a 2,500 mm/s datasheet figure produces an average of 700 mm/s to 1,100 mm/s in a real cell. Budget a 15 % to 25 % margin on the calculated total and you will rarely be wrong in the expensive direction.
The four-term move
A point-to-point move with trapezoidal velocity has a simple structure. With acceleration a, peak velocity v and distance d, the arm needs d_ramp = v² / a to accelerate and decelerate. If the move distance is shorter than that, the profile is triangular and peak speed is never reached.
Worked numbers for a mid-size arm with v = 2,000 mm/s and a = 8,000 mm/s²:
- Ramp distance: 2,000² / 8,000 = 500 mm. Any move shorter than 500 mm never touches rated speed.
- A 300 mm move is triangular. Peak velocity is the square root of (a x d) = the square root of 2,400,000 = about 1,549 mm/s, and the time is 2 x 1,549 / 8,000 = 0.39 s.
- A 1,200 mm move has 0.5 s of ramps covering 500 mm, plus 700 mm at 2,000 mm/s = 0.35 s, total 0.85 s.
- Add settling of 0.05 s to 0.25 s per stop where precision matters, more with a long or heavy tool.
A worked machine tending cycle
| Step | Distance | Type | Time |
|---|---|---|---|
| Home to infeed tray | 1,100 mm | blended | 0.80 s |
| Descend and grip blank | 120 mm | precise stop | 0.55 s |
| Gripper close | process | fixed | 0.30 s |
| Lift and traverse to machine | 1,600 mm | blended | 1.05 s |
| Wait for door open | process | fixed | 1.80 s |
| Enter, unload finished part | 700 mm | precise | 1.25 s |
| Index dual gripper | process | fixed | 0.45 s |
| Insert blank into chuck | 180 mm | precise | 0.90 s |
| Chuck clamp confirm | process | fixed | 0.60 s |
| Retract and clear door | 900 mm | blended | 0.70 s |
| Wait door close | process | fixed | 1.60 s |
| Place finished part on outfeed | 1,300 mm | blended plus stop | 1.15 s |
| Robot-attributable subtotal | 6.40 s | ||
| Fixed process subtotal | 4.75 s | ||
| Calculated total | 11.15 s | ||
| With 20 % margin | 13.4 s |
Read the two subtotals. Fixed process time is 43 % of this cycle, and no robot on the market shortens it. A faster arm improves the 6.40 s and leaves the 4.75 s untouched, so paying 30 % more for a 20 % faster machine buys about 1.3 s, or 10 % of the cycle. Reducing the door open and close time by 1 s would buy nearly as much for a fraction of the cost.
Five estimating mistakes
- Using rated speed on short moves. Below the ramp distance the arm never reaches it. Most machine tending moves are in that band.
- Ignoring payload derating. Acceleration is quoted at low load. At 80 % of rated payload expect 15 % to 30 % less.
- Forgetting settling. A long or heavy tool oscillates after a fast stop. The controller waits for it, and that wait is invisible in a simulation with default tool data.
- Counting only the robot. Doors, clamps, sensors, air valves and interlocks are usually a third to a half of the cycle.
- Trusting a demo video. Demonstrations run without a real workpiece, without safety-monitored slowdowns, and often without settling requirements.
Verifying before the money moves
Ask the vendor for a simulation with your part mass, your tool inertia and your real fixture positions, and ask for the reported figure to separate robot motion from process wait. Any credible integrator can produce that in a day. If the answer is a single number with no breakdown, the estimate has not been done, it has been asserted.
Then check the arithmetic yourself with the ramp distance formula. It takes ten minutes and it has settled more procurement arguments than any simulation screenshot.
Frequently asked questions
Why is my robot slower than its rated speed?
Because most moves are shorter than the acceleration ramp. With a rated 2,000 mm/s and 8,000 mm/s² acceleration, the arm needs 500 mm just to reach and leave peak speed, so a 300 mm move peaks at about 1,549 mm/s.
How much margin should I add to a calculated cycle?
15 % to 25 % for a well-understood cell, more if the part presentation is uncertain or if safety-monitored slowdowns will apply in part of the path. Margin covers settling, sensor debounce and the interlocks nobody listed.
Does a faster robot always shorten the cycle?
Only the motion portion. In typical machine tending, fixed process time is 30 % to 50 % of the cycle, so a 20 % faster arm often yields under 10 % on the total. Attacking door times and clamp times is usually cheaper.
What is blending and how much does it save?
Blending rounds the path through via points instead of stopping at each one, removing deceleration, settling and re-acceleration. On a six-point path it commonly saves 0.6 s to 1.5 s per cycle at the cost of a small path deviation.
How do I account for settling time?
Budget 0.05 s to 0.25 s per precise stop, at the higher end for long or heavy tools. Entering correct tool mass, centre of gravity and inertia into the controller lets it plan a profile that settles faster.
Sources
- ISO 9283, manipulating industrial robots, performance criteria and test methodsStandard cycle definitions and the conditions under which speed figures are quoted
- ISO 8373, robotics vocabularyInternational Organization for Standardization, definitions of pose, path and velocity terms
- World Robotics 2025, industrial robotsInternational Federation of Robotics, application mix used for the tending example