From Manual to Autonomous: How a 400,000 Sq Ft Distribution Center Cut Pick Times 35% With AMR Fleet Expansion

The warehouse was moving freight the way most still do: people pushing carts between bins, reading labels, scanning barcodes, walking. A lot of walking. The math was brutal. A single order pick in the 400,000 square foot facility took an average of 4.2 hours from receipt to dock. Half that time was pure movement. No value. Just distance.

The operation manager knew the constraint was not people or picking accuracy. It was travel time. The warehouse was organized into a grid, and pickers spent more time traversing the grid than selecting items. Adding more pickers made the grid more congested. The bottleneck was geometry, not labor.

Challenge

The facility processed 850 orders per day across three shifts. Peak season pushed that to 1,100. Hiring seasonal labor worked but came with training overhead, turnover friction, and wage pressure during peak windows. The operator faced a hard choice: build a second warehouse, or fix the one they had.

The core problem was not complex. The facility operated with a traditional pick-to-light and zone-picking model. Pickers worked their assigned zones and moved material to a central staging area. This meant long distances and significant congestion during peak hours. The logistics manager ran the numbers: average picker traveled 2.8 miles per 8-hour shift. Most of that was empty-handed travel back to their zone after delivering a cart.

Labor costs ran $1.87 per order. Peak season required hiring 40 additional workers. Turnover in those slots hit 35% annually. Training took two weeks.

Solution

The operator partnered with an AMR systems integrator and deployed a hybrid model. Sixty robots were introduced into the warehouse starting in Q4 2024. The system was not a full autonomous picking operation. Instead, robots handled the movement problem. Humans handled the picking problem.

Here is how it worked: A packer at a central station received an order. Instead of a picker walking the warehouse, the system dispatched a robot to retrieve a bin from a nearby shelf module and bring it to the packing station. The human selected the items. The robot returned the bin. The next order dispatched the next robot.

This inverted the geometry. Instead of people walking to bins, bins came to people. The robot fleet handled 70% of travel tasks. Pickers stayed at pack stations. Movement became predictable and measurable. The system coordinated robot traffic automatically, avoiding deadlock and collision through a fleet management platform.

Initial deployment was phased. Thirty robots launched in Q4 2024, handling single-shift validation. Thirty more came online in Q1 2025 across all three shifts.

Results

Order cycle time dropped from 4.2 hours to 2.7 hours. That is 35% improvement without adding facility square footage. Daily throughput increased from 850 to 1,039 orders per day at the same staffing level. During peak season, the operator reduced seasonal hiring from 40 workers to 14 workers. The labor cost per order fell to $1.21.

Annual labor cost avoidance totaled $2.1M. The payback period on the robot fleet and integration work was 18 months. The facility is profitable on that math alone. Training time dropped to three days. Turnover in permanent positions fell to 12% from 35% because the work became less physical and faster-paced.

Accuracy actually improved. Pickers were no longer fatigued from walking. Error rates dropped from 2.1% to 0.8%. Rework and customer returns declined accordingly.

The operator is now adding 30 more robots in Q3 2026 to handle continued demand growth without building a new facility. The second warehouse is off the table. That saves significant capital.

This is not a story about replacing people. This facility still employs 82 warehouse staff. It is a story about moving the bottleneck. When you remove distance as a constraint, throughput rises and people can focus on what they actually do well: selective judgment, problem-solving, and quality control. The robots handle the miles. Humans handle the decisions.