Key takeaways
- One master acceptance template with local annexes beats a different checklist at every site.
- Classify defects by safety, mission, and cosmetic tiers so rollout gates stay honest.
- Evidence packs should include logs, photos, and map clips reviewers can audit months later.
- Network and traffic tests belong in acceptance, not in the first week of unsupervised production.
- A vendor neutral integrator can run the same SAT format while adjusting only floor specific annexes.
What is multi site robot acceptance testing?
Robot acceptance testing across multiple sites is the structured proof that each building can run the same fleet standards on its own floors, networks, and traffic patterns. Site acceptance testing is not a single demo lap. It is a signed evidence pack that production rules, safety zones, and service escalation paths work before you clone the program elsewhere.
Corporate teams want comparability. Local managers want realism. The process holds both: a corporate master test script plus a local annex for dock lips, elevator handoffs, or clinic quiet hours that differ from the pilot building.
According to the International Federation of Robotics, worldwide sales of professional service robots grew 9 percent in 2024 to more than 199,000 units, while the robot-as-a-service fleet grew 31 percent to more than 24,500 units. Multi site operators are normal buyers now. Acceptance has to scale like the contracts do.
Why does a single site SAT fail on rollout two?
The pilot site often had the integrator on site daily, IT on speed dial, and a forgiving floor plate. Rollout two inherits none of that patience. If acceptance criteria were tribal knowledge, the second building either passes too easily or stalls over arguments about what good looks like.
Local floors change stopping distance, Wi-Fi handoff, and how pedestrians react. A scrubber that met coverage targets in a wide retail box may miss clinic corners or slow for narrow equipment aisles. Without local annex tests, corporate thinks the fleet failed when the map was wrong.
Transportation and logistics robots led professional service robot sales in 2024 with 102,900 units, a 14 percent increase, according to IFR data. Warehouses clone faster than hospitals, yet both need written gates before night one in a new building.
What belongs in the master test script?
The master script lists test categories every site must run: safety stop behavior, minimum coverage or move counts, charge and dock reliability, alarm escalation, and handoff to local staff. Each category defines pass metrics, tools, and required artifacts.
Keep scripts readable by role. EVS leads run cleaning sections. IT runs network roam tests. Security runs patrol camera and alert routing checks. Finance cares about signoff authority, not squeegee brands.
Version the master script when firmware or service tiers change. Sites mid rollout should not discover new criteria from a Slack message.
- Safety stops and zone compliance on live traffic samples
- Autonomous task counts versus agreed minimums
- Dock, charge, and water or payload cycles without intervention
- Ticket creation and remote triage response within SLA
- Staff handoff drill with named local owners
How should local annexes differ without breaking the standard?
Local annexes add tests tied to verified floor facts: ramp traction, elevator recovery, cross contamination rules between zones, or peak hour pedestrian density. They never remove a master safety test.
Cap annex length. Two pages of local checks beat twenty ad hoc notes from a walkthrough. Each annex item maps to a risk the corporate script did not cover.
Photograph and map annotate annex conditions. A loading dock lip that triggers slow zones in Tucson should not be rediscovered in El Paso.
How do you classify defects before signoff?
Use three tiers. Tier one is safety or regulatory stop ship. Tier two is mission impact that breaks agreed coverage or move counts. Tier three is cosmetic or logging noise that can ship with a dated fix.
Defect tiers drive rollout gates. Tier one blocks production. Tier two blocks handoff to local night crews until retest passes. Tier three enters a backlog with a visible due date.
IFR reported that robot-as-a-service in transportation and logistics rose 42 percent in 2024. Subscription vendors have incentive to close tier two defects fast because uptime ties to revenue. Put tier rules in the acceptance charter so disputes do not reopen at invoice time.
What evidence should each site archive?
An evidence pack includes exported logs for test windows, map screenshots with zone labels, short video or photo stills where privacy allows, and signed checklists with names and timestamps.
Store packs where corporate and local leads can retrieve them. Audits, insurance questions, and union discussions reuse the same files months later.
Name files by site, robot serial or pool ID, and test version. Future you should not grep a shared drive for final v3 FINAL.
Who signs off and what authority do they need?
Signoff is a small roster: local operations lead, corporate automation sponsor, integrator project lead, and IT when network tests ran. Each signs only their section.
Empower local leads to fail acceptance. If only corporate can say no, sites will pass unsafe configs to avoid delaying a regional opening.
Corporate retains authority to hold rollout until tier one and tier two retests pass. That balance keeps speed without normalizing shortcuts.
How do rollout gates sequence the next site?
Gate zero is master script approved. Gate one is pilot site accepted with a complete evidence pack. Gate two is each additional site accepted before production hours expand. Gate three is corporate audit sampling on a random site quarterly.
Parallel rollouts still need sequential acceptance per building. Sharing a pool of robots does not share a floor map.
Service Robot Co. runs turnkey robot deployment with the same SAT template across regions, adjusting annexes for local floors while keeping defect tiers and evidence formats identical. That is how multi site robot fleet management stays auditable when a new contractor joins mid program.
What mistakes extend acceptance for weeks?
Starting acceptance before maps and network baselines freeze wastes retests. Freeze baselines, then run scripts.
Letting sales sit in on pass fail calls pressures local teams to approve marginal runs.
Skipping night traffic because day tests looked fine. Most cleaning and AMR programs live at night.
Treating acceptance as training week. Training belongs in the plan, but it is not a substitute for measured pass criteria.
