How a high-volume automotive parts distribution facility deployed a mixed fleet of AGVs and AMRs to eliminate tugger-train bottlenecks across 12 assembly lines.
Background
A high-volume automotive parts distribution facility operating three shifts per day faced escalating internal logistics costs and increasing pressure to support faster production line replenishment cycles. With over 40 sub-assembly stations spread across a 45,000 m² floor, manual tugger trains were creating measurable bottlenecks.
The Challenge
Key pain points identified during the initial site assessment:
- **Cycle time variability**: Manual tugger routes averaged 28 minutes per circuit, with high variance (±9 minutes) due to traffic and driver availability.
- **Labor cost pressure**: Internal logistics represented approximately 14% of total facility labor cost.
- **Delivery accuracy**: Human routing occasionally resulted in part deliveries to incorrect stations, creating production stoppages.
Solution Architecture
The SG-D deployment consisted of:
- **16 laser-guided SG-D FMR units** assigned to primary pallet routes with fixed high-frequency loops between the receiving dock and main sub-assembly staging areas.
- **8 SG-D LMR units** handling small-parts replenishment and exception deliveries to stations with variable demand.
- **SG-D Fleet Control Center (FCC)** coordinating both fleets under a VDA5050-compliant traffic controller, with WMS integration for automated mission generation.
Magnetic tape fallback guidance was installed in three zones near heavy metallic presses where laser signal return was unreliable.
Results After 18 Months
| Metric | Before | After |
|---|---|---|
| Line-feed cycle time | 28 min | 15.7 min (−44%) |
| Cycle time variance | ±9 min | ±1.2 min |
| Logistics labor allocation | Baseline | −40% |
| Part delivery accuracy | 93.8% | 99.7% |
| Fleet unplanned downtime | 3.4 hrs/week | 0.5 hrs/week |
Key Lessons
**1. Integration planning must begin at project kickoff.** WMS integration was the longest lead-time item—starting it in month 1 rather than month 3 would have compressed the go-live timeline by approximately six weeks.
**2. Mixed-fleet architecture outperformed single-type deployments.** Early simulations suggested a homogeneous AGV fleet would be optimal. Operational data showed that the hybrid FMR/LMR model delivered 12% higher throughput than the modeled single-type scenario, because AMRs absorbed demand spikes that would have required AGV detours.
**3. Operator training is underestimated.** Shift leads needed structured training not just on the fleet management dashboard but on recognizing and correctly escalating edge-case robot behaviors. Facilities that invested in this training saw faster anomaly resolution and significantly lower unplanned intervention rates.