KOREA VIDEO REVIEW · COBOT TENDING
Cobots That Tend the Machine: A Walkthrough of Doosan Robotics' Machine-Tending Training
01What the two official videos are about
Doosan Robotics, spun off from the Doosan Group and publicly listed, is a leading player in the global collaborative-robot market. This official-channel "머신텐딩 application training" livestream runs over an hour, teaching from application concepts all the way to hands-on implementation—the tags even include FOCAS (FANUC's controller communication protocol) and DRL (Doosan Robot Language), at a depth aimed at integration engineers[1].
A second "field application case" video shows a real scenario: load/unload automation of a tooling system (툴링시스템)—the cobot loading and unloading against the machine[2]. A training course plus a field case is the most practical pairing of material for evaluating this kind of solution.
02The hard part isn't the arm, it's the integration
The structure of the training video is itself the answer—why spend an hour teaching a "pick it up, put it down, pick it up again" application? Because the engineering effort in machine tending lies outside the arm:
- Machine communication: the robot needs to know "is the cut finished, is the door open, is the chuck released"—this is exactly where protocols like FOCAS come in; without working communication, the robot can only wait dumbly or barge in[1].
- Gripper and part rack: the workpiece's shape, weight, and surface determine the gripper; the placement of incoming stock and finished parts determines the rack and positioning scheme.
- Safety and cycle time: a collaborative robot's "collaborative" is a safety grade, not a speed grade—scenarios with demanding cycle times need to be recalculated.
An old rule from the automation textbooks becomes concrete here: the bulk of an automation project's cost is integration engineering, not equipment procurement[3]. The vendor makes its training this deep precisely because integration capability decides whether adoption succeeds or fails.
03Who cobots suit: updated criteria
Cross-referencing the earlier related pieces in this series, the criteria for cobot tending can be organized into three layers:
| Condition | Traditional criterion | Updated |
|---|---|---|
| Batch size | Only worthwhile at high volume | The batch threshold is dropping—simpler setup and vision assistance make small batches viable (see the RoboJob piece) |
| Part type | Fixed parts first | Still holds, but AI vision is loosening this line |
| Floor space | Needs fencing and large space | Collaborative-grade needs no fencing (after risk assessment), so even old plants can fit one in |
The full "should you or shouldn't you" decision tree is in the intro to robotic machine loading—the criteria are moving, but the starting point of "first figure out exactly what you're automating" doesn't change.
04After tending automation: the next bottleneck
Readers who've made it to the fortieth piece in this series should be able to say this part themselves: once a cobot frees the "person who tends," the "waiting for a program" bottleneck floats to the surface. The robot loads and unloads parts day and night, machine throughput gets maxed out—and the supply rate of programs, tool lists, and fixturing schemes for each new order becomes the new ceiling (for the same conclusion in the European version see the Kostwein piece, and in the Japanese version the HILLTOP piece).
So the "tending automation" that the Korean video teaches, and the "prep automation" that BestAI CAM does, are two halves of the same blueprint: AI reads the 2D drawing, builds a 3D model, generates G-code according to the shop's tool library and machine limits, with built-in cutting simulation and independent AI dimensional cross-verification (see the complete guide to automated CNC programming)—the cobot tends the machine, the AI tends the program, and people tend the judgment. With all three in place, "lean staffing" stops being a slogan.
05FAQ
What is the most underestimated cost in cobot machine tending?
Integration engineering: machine communication (such as FANUC's FOCAS protocol—the robot needs to know when a cut is finished, the door is open, and the chuck state), gripper and part-rack design, safety risk assessment, and cycle-time tuning. The arm itself is only part of the cost; the fact that the official training video spends an hour on integration shows exactly where the engineering effort lies.
Are high-mix, low-volume shops suited to cobot tending?
The barrier is falling: collaborative-grade means no fencing, setup tools are simpler, and AI vision now enables changeovers without teaching. But you still have to cost the changeover—what a person has to do at each batch change, and how long it takes. We recommend taking your three most typical parts and asking a vendor to run the actual changeover process, then deciding against your batch structure.
After adding a tending robot, why does the machine often sit idle?
The common cause is a shifted bottleneck: tending automation raises machine throughput, but the supply rate of programs and process prep can't keep up, so the machine ends up "waiting for a program." The fix is to automate the prep chain in parallel—AI reads the drawing, models, generates code to the machine's limits, and verifies by simulation—so the program supply rate keeps pace with the robot's feeding rate.
We'll let you know when new articles and video walkthroughs go live—no inbox flooding, one-click unsubscribe. (newsletter in Chinese)
TEND THE MACHINE, FEED THE PROGRAM
The cobot tends the machine, the AI tends the program—the full lean-staffing picture
Before you evaluate a tending robot, measure your program supply rate: AI from drawing to machine-ready program, demonstrated on one of your drawings.
Talk to an onboarding advisor Training courses06References
- DOOSAN ROBOTICS (YouTube). 머신텐딩 어플리케이션 교육, 두산로보틱스 (Kor) (livestreamed 2021-11-04). youtube.com/watch?v=rnnVCgbSuto
- DOOSAN ROBOTICS (YouTube). [현장 적용 사례] 머신텐딩 - 툴링시스템 로딩&언로딩. youtube.com/watch?v=j5bMSknpY3o
- Groover, M. P. (2019). Automation, Production Systems, and Computer-Integrated Manufacturing (5th ed.). Pearson.
- Monostori, L., Kádár, B., Bauernhansl, T., Kondoh, S., Kumara, S., Reinhart, G., Sauer, O., Schuh, G., Sihn, W., & Ueda, K. (2016). Cyber-physical systems in manufacturing. CIRP Annals, 65(2), 621–641.