Three Families of Robotics Goals and How Each One Drifts

Robots do not care about your goals. A conveyor-feeding arm will keep placing parts the same way whether the target on the wall still makes sense or not. The people around it, however, set targets, and those targets age. A robotics operations team in a warehouse, a factory cell or a lab can hold dozens of written goals, from cycle time to maintenance routines to training plans. This comparison looks at three families of goals found in most automation teams, shows how each one tends to go out of date and explains what a monthly reread of the written list can catch in each case.

The aim is analytical, not prescriptive. For each family I describe what drifts, what evidence is useful at the reread and what the typical decision looks like.

Family one: throughput and performance goals

These are the goals everyone writes first: parts per hour, pallets moved per shift, picks per minute. They are concrete, measurable and tied to money, which makes them feel permanent. In practice they drift for three reasons. The product mix changes, so a rate that was reasonable for one item is not for another. Upstream or downstream processes shift, making a local maximum pointless. And equipment ages or is modified, so the original baseline no longer describes the machine.

Evidence for the reread: a recent sample of real production, not a best-case test. Compare the written target with what the line produced across a typical week, including changeovers and short stops.

Typical decision: reshape the goal. Often the number needs to be conditional, for instance by product family, instead of a single figure for everything. For readers who like a refresher on how factory teams think about indicators, the article on key performance indicators that matter for robotic cells is a reasonable companion.

Family two: reliability and maintenance goals

These goals concern keeping the system running: scheduled inspections, spare-parts holdings, response routines after a fault. They drift more slowly than throughput goals but in a subtler way. A goal that says “inspect every gripper weekly” may have been written when one gripper type was in use. A year later, three types are in service and the inspection routine covers only the oldest.

Evidence for the reread: the maintenance log and the list of equipment currently in service. Place them side by side and ask what the goal fails to cover.

Typical decision: widen or narrow the goal to match what exists today. Sometimes the right move is to retire a goal about equipment that has been decommissioned, which sounds trivial but is surprisingly common.

Family three: skills and people goals

The third family concerns the team: who is trained on which cell, who can handle a fault alone, who is learning a new programming environment. These goals drift because the technology changes under the team’s feet. A skill goal tied to one controller platform loses urgency if the plant starts moving toward another.

Evidence for the reread: the roadmap of equipment changes over the next few months, set beside each person’s learning aims.

Typical decision: redirect. The person’s aim to learn stays, but the subject moves. Some teams also use the reread to notice where only one person holds a skill, and turn that observation into a pairing goal.

When people on the team also study outside their own plant’s tools, they may browse other fields for inspiration. One neutral example is AirCrewU, where people in aviation can take courses online and which presents itself as independent. It has nothing to do with robotic cells, yet it shows how a specialist industry structures study, which prompts the useful question of how your own team would label its skill areas.

Why evidence beats memory in this setting

Robotics teams often rely on memory of how a line behaved last month, and that memory is skewed by the loudest event. A single dramatic fault can make a whole month feel unreliable, while a slow decline in cycle time goes unnoticed because no one incident marks it. Pulling even a small sample of logged data into the reread guards against both distortions. The sample need not be elaborate: a week of production counts, the fault list and a note on what changed in the cell are enough to ground the conversation.

It also helps to write down, before looking at the data, what you expect to find. If the written goal says output should be steady and the sample shows a decline, the surprise itself is information. Teams that record their expectation first tend to learn more from each reread, because they can see where their mental model of the cell differs from the cell itself.

Side by side

  • Speed of drift: throughput goals move fastest, skills goals next, reliability goals slowest.
  • Best evidence: production samples for throughput, logs and equipment lists for reliability, roadmaps for skills.
  • Common fix: condition the number, update the scope, redirect the subject.
  • Risk if ignored: chasing a target that no longer matters, missing equipment the routine does not cover, training for a platform nobody will use.

A short worked case

Take a warehouse team operating a fleet of autonomous carts. Their sheet has three goals, one from each family. The throughput goal says to move a certain volume per shift. The reliability goal calls for a weekly battery and sensor check. The skills goal says that every technician should be able to recover a stuck cart.

At the monthly reread, the team compares each against evidence. The throughput target was written before a new, heavier product line was introduced, and the carts now spend more time on longer routes. The team decides to split the goal into two product-based figures. The weekly check, they discover, covers the first generation of carts only; a newer batch has different sensors. The goal is widened to cover both. And the skills goal has a gap: only two technicians have ever recovered a stuck cart, so the goal is reworded into a pairing plan for the next quarter. For general context on how such fleets fit into a warehouse, a full overview of AGV automation in warehouses is worth a read.

Choosing between a global reread and a per-family reread

Some teams ask whether to review everything together or one family per month. Each approach has a cost. A single combined reread keeps all the goals in view and reveals conflicts, such as a throughput target that leaves no room for the maintenance goal. A rotating review gives each family more attention but risks letting one family go three months unchecked, which is too long for the fastest-drifting goals.

A compromise that many teams find workable: reread everything monthly in a short session, and take a deeper look at one family each month on a rotating basis. The short session catches drift; the deeper look catches subtler issues.

What good evidence does and does not do

Evidence helps you decide, but it does not decide for you. A drop in output may mean the target is wrong, or it may mean a fault needs repair. The reread is the moment to ask which. If a goal has been missed for two months running and the cause is a known, fixable problem, the goal stays and the fix gets scheduled. If the cause is a change in what the work now is, the goal changes.

Practical steps for your team

  1. Sort your current goals into the three families.
  2. Attach one piece of recent evidence to each.
  3. Hold a thirty-minute session and decide each goal’s fate with a one-line reason.
  4. Record the outcome where the whole team can see it.
  5. Choose one family for a deeper look next month.

Machines repeat. Goals should not. A short, regular check of the written targets keeps a robotics operation aimed at what the work needs this month, not at what it needed when the sheet was first drafted.

Related Posts

Eight Things a Finished Robot Cell Teaches You About Operations

A technician reads a finished pick-and-place cell against its commissioning binder. Eight structural lessons for anyone running robotics operations.

Read More

Robot Cell Fault Messages: Vendor Style Against Plain Words

The same robot fault written twice: once as controller shorthand and once for the operator at the screen. A practical side-by-side.

Read More

Robotics Operations on a Short Parts List: A Beginner Primer

A precise beginner primer on why robot builders and operators benefit from limits on sensors, parts and behaviours, with a glossary of the terms you will meet.

Read More

Commissioning a Robot Fleet Slowly, Step by Step

A warehouse team learned that bringing robots up at walking speed, one check at a time, finished the job sooner than a fast start. Here are the steps.

Read More

Robotics Operations and the Done List: Practical Answers

What belongs on a done list for robots, how it differs from the maintenance schedule, and who writes it? Direct answers for teams running automated equipment.

Read More

Fallback Skills That Robotics Operations Employers Look For

Robotics operations rewards people who know what happens when a step fails. Seven fallback skills to build, and how to turn them into a career plan of your own.

Read More

Leave a Reply