ZhilianhaiHow to Run a Robotics Pilot That Survives the First 90 Days
Start with a task, not a technology
A pilot that begins with “we should try a robot” almost always ends with a machine in a corner. Start from one task that is repetitive, has a stable input, and can be measured today with a stopwatch. If the task cannot be described in one sentence with a start signal and an end signal, it is not ready for a pilot.
Write the success metrics before installation
- Cycle time per unit, measured over a full shift, not a demo run.
- First-pass yield and how often a human has to intervene.
- Availability across at least 20 production days, including tool changes.
- Operator time saved per shift, and the time the operator now spends supervising.
Define the exit criteria
A pilot needs three endings: success (scale up), failure (stop and document), and inconclusive (extend once with a written reason). Fix those thresholds up front so the decision is not made by whoever is most enthusiastic in week nine.
| Checkpoint | What to review | Owner |
|---|---|---|
| Day 1-10 | Installation, safety check, dry runs | Integrator + plant EHS |
| Day 11-45 | Shifts running without rescue, cycle-time data | Production lead |
| Day 46-90 | Maintenance load, operator feedback, cost per unit | Project sponsor |
Keep a dated log of every intervention. It becomes the business case for scaling — or the evidence that stops a bad investment early.