Why Your Robot Arm Doesn’t Land Exactly Where You Told It To — ARTUS
Precision Guide

Why doesn’t your robot arm land exactly where you told it to?

Five numbers on a joint module spec sheet quietly decide how precise a robot really is. Here’s what each one means, in plain English.

Every robot joint has a motor, a gearbox, an encoder and bearings working together — and none of them are perfect. The gap between “where you told it to go” and “where it actually stops” gets measured five different ways, and each one tells you something different about how the joint will behave.

01

Backlash

In plain terms: the tiny bit of “play” in the gears. Spin a joint one way, then reverse — there’s a brief moment where nothing happens yet, because the gears are still taking up slack before the output actually starts moving back.

Why it matters: every direction change causes a small delay. For tasks that reverse a lot — tracing a contour, drawing a shape — that delay shows up as a wobble at each turn.
gap = backlash turning forward reverse: brief pause first
drive gear mating gear
02

Lost Motion

In plain terms: send the joint to the exact same spot from two different directions — once approaching from the left, once from the right — and it usually won’t land in exactly the same place. That difference is lost motion.

Why it matters: backlash is one cause, but not the only one — bending, friction and gear flex add to it too. It gives a fuller picture of consistency, which matters for precision assembly or inspection.
SAME TARGET approach from left approach from right ↑ two slightly different landing spots
03

Transmission Error

In plain terms: even while spinning smoothly with no reversing involved, the output almost never rotates in perfectly even steps. Tiny manufacturing imperfections make it speed up and slow down slightly, over and over, through each rotation.

Why it matters: this shows up as a faint ripple in speed and position throughout the rotation — not just at direction changes. For smooth continuous motion, like drawing a curve, that ripple is the difference between a clean line and a slightly shaky one.
ideal, perfectly even rotation actual rotation — tiny repeating ripple
04

Positioning Accuracy

In plain terms: how close the joint actually gets to the exact spot you commanded, measured against a fixed, known reference point.

Why it matters: a joint can be very consistent and still be a little off target every time — like a dartboard where every throw clusters tightly, but off to one side. That’s an accuracy problem, and it’s different from a consistency problem.
target center tight cluster — but off-center = poor accuracy, good repeatability
05

Repeatability

In plain terms: how tightly the joint returns to the same spot, over and over, under the same conditions — regardless of whether that spot is the “correct” one.

Why it matters: most factory automation cares about this more than absolute accuracy. Once a robot is taught a position for pick-and-place or dispensing, it just needs to come back to the same exact spot every cycle.
target center tight cluster, right on target = good accuracy and repeatability
Quick Reference

All five, side by side

SpecWhat it’s really askingWatch this if you’re doing…
BacklashHow much slack before direction reversal takes effectFrequent direction changes
Lost MotionDifference in landing spot from opposite approach directionsBidirectional precision tasks
Transmission ErrorRipple in motion during continuous rotationSmooth, continuous trajectories
Positioning AccuracyCloseness to the exact commanded positionWorking to fixed external coordinates
RepeatabilityConsistency returning to the same taught positionRepetitive pick-and-place, dispensing
Reversing often

Frequent direction changes

Prioritize low backlash and low lost motion — they control how the joint behaves the instant it changes course.

Smooth paths

Continuous trajectories

Transmission error, encoder quality and calibration matter most for keeping motion even throughout.

Exact coordinates

Absolute positioning

Look at positioning accuracy — and evaluate it at the whole-system level, not just the gearbox.

Same task, repeated

Repetitive automation

Repeatability is usually the number that matters most for pick-and-place and assembly lines.

None of these five specs tells the whole story by itself. A joint can be excellent on paper for one and mediocre for another — which is exactly why it’s worth understanding what each one actually measures before comparing spec sheets.

The right joint module is the one matched to your motion pattern — not just the one with the smallest number on a single line.

Not sure which spec matters for your application?

Tell us your torque, speed, payload and positioning requirements — we’ll help you match the right joint platform.

Request Technical Evaluation →