Autonomous Forklift Pilot Program Guide
Most autonomous forklift programs start with a pilot — one or two trucks running a defined task for 3-12 months as a proof of concept. Done well, the pilot builds organisational confidence, surfaces integration issues early, and produces the operating data that drives the full-scale business case. Done poorly, the pilot becomes a $300k experiment that goes nowhere because nobody defined what success would look like before it started. This guide is about doing it well.
The Five Pilot Questions
Before signing the pilot purchase order, answer these five questions in writing:
- What task is the pilot doing? One specific repetitive flow — not "let's see what the trucks can do."
- What does success look like, quantitatively? Specific KPIs with measurement methodology defined upfront.
- What's the exit criterion that triggers full-scale rollout? "We'll discuss it at the end" is not an exit criterion.
- What's the off-ramp if it doesn't work? Buy-out terms, redeployment options, lessons-learned process.
- Who owns the pilot? Single named accountable executive — not a committee.
Pilot Scope: Pick One Task, Do It Well
| Good Pilot Scope | Bad Pilot Scope |
|---|---|
| "Replenish 50 ground-level pick locations from reserve floor stock, 6 days/week" | "Run the warehouse autonomously" |
| "Move pallets from receiving dock 3 to staging area B" | "Help the team with whatever's needed" |
| "Cross-dock pallets between bays 1-4 for inbound/outbound flow" | "Increase efficiency" |
| "Replace 1 manual reach truck on overnight shift" | "Replace forklift drivers" |
Narrow scope works. Broad scope fails — the autonomous trucks struggle with edge cases, the team gets frustrated with limitations, and the program loses momentum.
Pilot Success Metrics
Throughput
Pallets per hour, cycle time per task, peak-period throughput. Compare directly to baseline manual performance over the same period; no apples-to-oranges comparisons.
Availability / Uptime
% of scheduled operating time the truck was actually productive. Below 90% in pilot is concerning; below 80% suggests integration or fit issues.
Safety Performance
Zero incidents is the baseline expectation. Track near-misses and operator reports of unexpected behaviour as well; these flag early warning signs.
Operator Acceptance
Survey-based: do warehouse operators consider the autonomous truck a help or a hindrance? Operator buy-in matters as much as throughput data for full-scale rollout.
Integration Fit
Does the WMS integration deliver the data accuracy required? Are exception flows handled or do they fall through to manual processing? This is where many pilots reveal hidden gaps.
TCO Validation
Did the actual operating costs match the projected business case? Energy, maintenance, labour displacement — all measured and validated against assumptions.
Pilot Duration
The right pilot length depends on what you're trying to learn:
- 3 months — minimum viable pilot for stable repetitive task; enough to validate basic operation
- 6 months — standard pilot length; covers seasonality and surface most edge cases
- 12 months — required for sites with significant seasonal variation (vintage, retail peak) where summer and winter operating profiles differ
Common Pilot Pitfalls
- Comparing to perfect manual operations — pilot trucks compared to a hypothetical zero-defect manual operation that doesn't actually exist
- Insufficient training — pilot operators not trained on the autonomous workflow, leading to operator-induced "failures" attributed to the truck
- Scope creep — "while it's here, can we also try X?" leads to unfocused operating data and unclear conclusions
- No baseline data — nobody measured the manual operation properly before the pilot started, so there's nothing to compare to
- Vendor over-promise — vendor claims of capability that don't survive contact with real operating conditions
- Pilot becomes permanent — pilot runs for 4 years without exit decision because nobody owns the next decision
Recommended Pilot Configurations
| Pilot Use Case | Recommended Pilot Fleet |
|---|---|
| Cross-dock pallet flow (1 dock pair) | 1 × 3.0T Counterbalance |
| Pick-face replenishment | 1 × 1.5T Pallet Stacker |
| High-bay storage automation | 1 × 2.0T Reach Truck |
| Manufacturing line-feeding | 1 × 4.0T Tractor |
| Bulk pallet ground transport | 1 × 2.0T Pallet Mover |
Robots Now! Pilot Approach
We support pilots with structured KPI definition (jointly agreed upfront), weekly operating reviews during pilot, monthly executive reviews, and a documented exit gate that converts pilot success into rolling-stock orders. We don't run "see how it goes" pilots; the pilot has a destination, and we make sure we get there.