Columbus DC Lifts AMR Fleet Uptime to 99.5% With Maintenance Scheduled Around Charging

A fleet that stopped too often to hit its numbers

A 480,000-sq.-ft. fast-moving consumer goods distribution center in Columbus, Ohio, runs three shifts across 20 operating hours a day, with 45 autonomous mobile robots moving orders through the building. Peak demand reached 22,000 order lines a day, against an average of 14,000.

Before the program, fleet uptime sat at 93.1 percent, and the site logged 6.8 unplanned stoppages a week. The vendor-published source estimates each stoppage cost $4,100 in lost throughput. Recovery averaged 2.3 hours, and night shift recovery ran longer, at 3.1 hours.

The source also describes a tooling problem. The native maintenance tools for the robot platforms operated in separate data silos, which kept fleet health from being read in one place.

Predictive health scoring tied to the charging window

The fix was less about new hardware and more about when and how the team worked on it. Each robot received a predictive health score, and work orders were scheduled into the 4-hour nightly charging window so repairs did not compete with live orders.

The source reports integration live within 72 hours, health scores reliable by week 6, and full predictive operations by weeks 9 to 12. Battery and sensor work followed the same logic, with planned replacements instead of emergency swaps.

• Score each robot's health continuously and flag drift before a fault.
• Book work orders into the 4-hour nightly charging window.
• Replace seven battery packs during charging windows, guided by cell-voltage spread monitoring.
• Calibrate lidar on a schedule rather than after a navigation fault.
• Set wheel replacement individually, between 620,000 and 1,100,000 cycles, instead of a uniform 800,000.

What the vendor reports, and how far to trust it

Fleet uptime rose from 93.1 percent to 99.5 percent, and unplanned stoppages fell from 6.8 to 0.4 per week, a 94 percent reduction. Night shift recovery dropped from 3.1 hours to 1.4 hours, a 55 percent improvement. The source also reports zero unplanned mid-shift stoppages across 16 consecutive weeks.

The $1.4 million throughput figure is the vendor's headline. Its own inputs, 6.4 fewer stoppages multiplied by $4,100 across 52 weeks, come to $1,365,000. The source says labor and parts savings are included but does not itemize them, so the total should be read as a vendor estimate. The 5.2x twelve-month ROI carries the same caveat.

Two checks matter before anyone quotes these numbers. The study period is described as both an 8-month period and a 12-week baseline while the ROI is labeled 12-month. The remaining 0.5 percent of downtime is described as planned maintenance, which means the uptime metric may include planned time. As context, a 2025 peer-reviewed review lists a system availability target of 97.5 percent, from a baseline of 94.2 percent, which offers a useful external yardstick for a gain of this size.

What this means for a fleet owner planning a lease

This is a documented real-world example we analyze, not a client of Service Robot Co., and we did not run this deployment. The lesson travels well, though. Uptime depends on battery health, charging schedules, sensor care and one record of what each robot has done, and those pieces often sit with different vendors.

Service Robot Co. is a full-service, OEM-neutral commercial robot integrator for US businesses. We select robots across manufacturers, then finance, deploy, integrate, train and service every unit through a nationwide US engineer network. Customers lease the robots on a 3-year lease, with one fixed payment quoted after a site assessment, so the maintenance plan and the robot leasing terms come from the same team.