Roaming cobot stacks trays and picks conveyor parts at a 54 person molder

Custom injection molding, over 1,300 different parts for more than 100 customers · Pick and place · Vendor published

The job

At AIM Processing in Colorado, a worker sat at the palletizing station and every 90 seconds moved a tray into position for a cartesian robot, spending only a couple of seconds actually working each cycle. Because the job was high mix and low volume, no single application justified a dedicated caged cell, so the station stayed manual and ran only about 40 hours a week. A UR5e now picks and places the trays, reading a QR code on each tray with a wrist camera to determine orientation and placing it to 0.5 mm so the cartesian robot can palletize small parts into the compartments. The same cobot is moved to a second application picking parts off a conveyor and boxing them, and to injection molding machine tending.

The shop

54 employees, published-by-vendor. The tray station went from 40 hours a week to 24/7 operation, published-by-vendor. Shift pattern beyond those two figures: not published.

The equipment

Universal Robots UR5e on a mobile platform. Robotiq Wrist Camera for QR code reading and conveyor part location, Robotiq 2F-140 gripper, and a pneumatic gripper designed in house. Flexxbotics Flexx Reference lockout assembly, a 6 degree mechanical mating system, so the arm returns to a known origin at each station. Existing cartesian palletizing robot, conveyor with a proximity sensor, and a gate link relay box passing signals through the UR5e 24 volt I/O. No external PLC.

The published numbers

Output productivity: fourfold, stated as a 400 percent increase, published-by-vendor. Payback: ROI inside 15 weeks, published-by-vendor, with the owner stating the robot had paid for itself after the fifth run of the first application. Prior manual cycle: a tray move every 90 seconds taking a couple of seconds of operator time, published-by-vendor. Placement accuracy required: 0.5 mm, published-by-vendor. Operating hours: from 40 hours per week to 24/7, published-by-vendor. Applications launched since: more than a dozen, published-by-vendor. Headcount change: not published. Scrap: not published. Cell cost: not published. Cobot cycle time: not published.

Where a source did not publish a figure, this entry says so rather than guessing.

What we would check before believing it applies to you

The engineering content here is not the cobot, it is the mechanical mating plate that lets a cart mounted arm return to a known origin so stored programs still land within 0.5 mm without reteaching waypoints. A shop chasing this should verify its own stack of floor, cart and fixture actually supports that, because a mobile arm inherits every bit of slop underneath it, and a redeploy that needs two hours of reteaching destroys the whole premise. Read the 400 percent as an availability gain, not a speed gain: it comes from running 24/7 instead of 40 hours a week, so a plant already on three shifts will see almost none of it. The 15 week ROI is the owner's own arithmetic on one application counted over five runs, and nothing published tells you whether it includes the camera, grippers, mating hardware or engineering time. Note also that the cobot is only handling trays; a separate cartesian robot does the actual part picking, which is a far harder problem.

Sources