What changed when they automated
What changed when a 25 person optics shop ran its Haas overnight with a vision guided cobot
Walt Machine of Lumberton, Mississippi gets one order a spring for about 6,000 camera housings and ran it on a single Haas with the machine idle all night. A UR10 with a wrist camera now loads blanks from a table. Daily output doubled, and the published account says plainly that cycle time did not improve at all.
The short answer
A 25 person shop doubled the parts it machined per day by letting a cobot run the machine four to six extra hours a night, with a part that takes 30 to 45 minutes a side. The account below is the maker's own case story, so read every number as the vendor published it, not as an audited result. The maker's account is unusually honest on one point: the robot did not make the cycle faster, it made the job finish sooner.
What the job looked like before
Each spring the shop receives an order for about 6,000 precision aluminum camera housings and runs the whole thing on one Haas VM-3 vertical machining center. Working eight hour days, the order took several weeks and the machine sat idle every night. Hiring a temporary operator did not help, because the order would be nearly finished by the time the person was trained.
What they put in
A Universal Robots UR10 with a Robotiq wrist camera mounted above a Robotiq two finger gripper. The operator sets 15 to 20 unmachined blanks on a table. The robot photographs the table, takes a detailed shot of one part, loads it into the vise and signals the control to cycle. The president reports that teaching the camera a new part takes about 10 minutes. In a later vendor follow up the shop added robots and hung a UR10 from a ceiling rail to tend three machines in a small footprint.
What changed, in their numbers
Twice as many parts machined per day with the first robot, published by the vendor. An extra four to six hours of unattended running per night. About 6,000 housings inside a two month window. Teaching time of about 10 minutes per part. The later expansion is described as a 2,500 percent production increase, a vendor figure that blends added robot capacity with a large jump in customer demand. Payback, scrap and cell cost were not published.
What this does not tell you
It does not tell you what the cell cost or how fast it paid back. The vision task here is trivial, a flat blank on an uncluttered table under shop lighting, so it says little about parts that arrive nested, oily or overlapping. And the 2,500 percent figure should be read with real caution, because it comes from a later vendor follow up and is not a per hour speed claim.
What to verify in your own plant
Whether your cycle is long enough that one blank load buys a real unattended stretch. Whether your parts can be presented on a table in a known way without custom trays. Whether you have overhead clearance if floor space is the constraint, since a rail was the transferable idea here. And whether the machine has, or can take, an automatic door.
Sources: https://www.mmsonline.com/articles/vision-guided-cobot-doubles-daily-production and https://robotiq.com/resource-center/case-studies/walt-machine
Describe your machine and your parts to Joe, the concierge on this site, if you want a preliminary read on whether unattended hours are there to be had. It is free and it asks about the door and the part presentation first.
Related on this site
- CNC machine tending
- Pick and place
- Machine tending in a small shop wasn't automatable until the robot could move between machines
- Robots for machine shops
- The casebook of published deployments
Automation here bought machine hours, not speed. A 30 to 45 minute cycle per side is what made one table of blanks worth four to six unattended hours.
Put it to the test on your own job
Tell Joe what you are running and he will apply this to your numbers, including telling you when a robot is the wrong answer.
More notes