Cobot Machine Tending: What You Need to Know Before You Start

Feasibility checklist

Work through these questions before committing to a cobot tending project:

If you have more red answers than green, the application needs redesign before integration — not a different robot. The most common project cancellation reason is discovering a red-answer problem after the integrator has been engaged and preliminary design has begun.

Cobot selection for machine tending

The right cobot for machine tending is determined by three parameters: reach, payload, and machine interface compatibility.

Reach

The cobot must reach from its mounting position to: (a) the part staging area, (b) the machine door opening, and (c) the chuck or fixture. Measure these distances carefully — the working radius must include the cobot base offset, not just the target points. A common mistake: the cobot can reach the chuck but cannot reach the staging table while staying within safe operating limits. Use the robot manufacturer's reach diagrams, not just the nominal reach specification.

Payload

Calculate: part mass + EOAT mass + any cable drag at full extension. The EOAT (gripper + mounting plate + sensors) typically weighs 0.5–3 kg. A 5 kg part with a 2 kg EOAT requires a minimum 7 kg payload robot — a UR10e (10 kg) or FANUC CRX-10iA (10 kg) provides comfortable margin. Do not undersize the payload — operating at the rated limit reduces robot life and increases faulting rate.

CNC machine interface compatibility

Not all cobots integrate equally well with all CNC machines. Universal Robots has the largest ecosystem of CNC-specific URCaps (machine interface plug-ins) covering FANUC, Siemens, Haas, Mazak, Okuma, and DMG MORI. FANUC CRX robots integrate natively with FANUC CNC controllers (PMC/M-code). Check whether your CNC manufacturer's machine interface plug-in exists for your chosen cobot brand — it will save weeks of custom integration work.

EOAT for machine tending

End-of-arm tooling is the most application-specific component and the most common cause of delays and rework in machine tending installations.

Double-gripper design

The most productive EOAT for machine tending is a double gripper — two sets of jaws on the same EOAT, one for the raw part and one for the finished part. The sequence: robot picks raw part in gripper A → enters machine → gripper B picks finished part from chuck → robot rotates 180° → gripper A loads raw part into chuck. This eliminates one approach-retract cycle per part and can reduce load/unload time by 30–40%.

Gripper type selection

• Pneumatic parallel gripper: Standard choice for round or rectangular parts with consistent geometry. Fast, reliable, low cost. Requires compressed air.
• 3-jaw gripper: Better for round parts (self-centres without a precision part-presentation nest). More expensive but eliminates part presentation accuracy requirements.
• Vacuum gripper: Suitable for flat raw stock (sheet metal, bar stock with flat end). Not suitable for oily machined finished parts.
• Custom machined fingers: Necessary for complex part geometry. Budget $2,000–$8,000 for custom finger design and machining. Allows tightest part grip accuracy.

Coolant and chip considerations

Machined finished parts may be wet with cutting coolant. Coolant on a finished part surface compromises vacuum gripper performance — use mechanical grippers for finished-part removal. Chips on the chuck face or staging area can cause part mis-seating — consider an air blast (automated or manual at changeover) to clear chips before loading. Some EOAT designs incorporate a small air blast nozzle.

Part staging requirements

The staging system — how raw parts are presented to the cobot and how finished parts are stored — determines the maximum unattended run time. This is often underinvested in first installations.

Raw part staging options

• Pallet / tray system: Parts loaded into a fixture tray with defined pocket locations. The cobot picks each part from a known XY position. Tray capacity determines run time — a 40-part tray at 5-minute cycle time = 200 minutes of unattended operation.
• Stack/magazine feeder: Bar stock stacked in a magazine; the cobot picks from the top of the stack. Simple but limited to one part type at a time.
• Vibratory bowl feeder: For small parts requiring orientation; adds $5,000–$20,000 to cell cost but enables truly automated raw stock handling for the right part geometries.
• Vision-guided bin picking: Robot with 3D vision picks randomly oriented parts from a bin. Highest flexibility but highest cost ($20,000–$40,000 in vision system cost) and longest commissioning time. Not recommended for a first installation.

Common integration mistakes (and how to avoid them)

1. Undersizing the staging capacity

The most common reason cobot tending fails to deliver the expected ROI: a 20-part tray that empties every 90 minutes, requiring the operator to reload it constantly. The operator ends up spending more time at the machine, not less. Rule of thumb: design for at least 3–4 hours of unattended operation per operator visit. If your cycle time is 5 minutes, you need a 36–48 part tray minimum.

2. Not automating the CNC door

Shops with manual CNC doors sometimes try to integrate a cobot without automating the door, using creative workarounds (leaving the door open, using the cobot arm to push the door). These workarounds fail in production. Budget $500–$2,000 for a proper pneumatic door actuator — it is one of the best money-to-ROI investments in the cell.

3. Ignoring chip management

In long unattended runs, chips accumulate on the staging area, the chuck face, and in the machine. A chip buildup on the chuck face causes the next part to seat incorrectly, producing scrap. In lights-out operation (overnight), chip management must be planned: chip conveyor, in-machine chip conveyor, or shortened unattended run duration that doesn't exceed the chip management capacity.

4. Programming for a single part family, forgetting about changeover

The cobot program works perfectly for part A. Six weeks later, production switches to part B. The integrator hasn't been retained and your team doesn't know how to modify the program. Build internal cobot programming competency from day one — attend the training, document every program, and establish a process for part changeovers. UR's teach-by-demonstration interface is specifically designed for operators to manage this themselves.

5. Incorrect payload calculation

The cobot's rated payload is the maximum mass at the wrist flange, measured at full extension and worst-case orientation. EOAT designers sometimes give the EOAT weight without accounting for the moment load — the effective payload at the tool tip with the EOAT mass offset from the wrist is higher than the static weight. Use the robot manufacturer's payload calculator (UR has a free tool) with the correct centre-of-gravity offset before finalising EOAT design.

High-mix machine tending

High-mix (many different part numbers, short runs) is the hardest cobot machine tending scenario. The challenges:

• Many different programs required; operators must know which to select
• EOAT changeover required for different part geometries — can take 10–30 minutes if not designed for quick change
• Staging fixture changeover for different part trays
• Shorter runs mean higher programming time as a fraction of production time

Solutions for high-mix:

• Flexible gripper: Electric parallel gripper with wide stroke range (e.g., Robotiq 2F-140 or OnRobot 2FG7) that handles multiple part diameters without jaw changes
• Quick-change tool coupling: Allows operator to swap EOAT in under 2 minutes without tools
• Universal staging tray: Configurable tray with adjustable locating pins that accommodates multiple part sizes
• Vision system: Camera over staging area recognises part type and selects the correct program automatically

High-mix cobot tending works best when: runs are at least 10 pieces (enough to justify setup time), parts share similar geometry (cylindrical or prismatic) allowing one flexible gripper, and operators are trained to manage changeovers confidently.