Automation & Robotics — Engineering Decision Guide

Use a 6-axis articulated arm for flexible, multi-orientation tasks (welding, machine tending, assembly). Use a SCARA for fast, rigid, planar pick-and-place and vertical insertion. Use a delta for very high-speed lightweight picking (packaging, sorting). Use a Cartesian/gantry for large work envelopes and simple linear moves. Use a collaborative robot (cobot) where humans share the space and speed/payload are modest.

Robot geometry sets reach envelope, speed, payload, and orientation freedom. A 6-axis arm reaches any pose within its envelope — maximum flexibility. SCARA trades orientation freedom for rigidity and speed in a plane. Delta's parallel linkage gives extreme speed at low payload. Cartesian scales the envelope cheaply for straight-line work. Cobots trade speed and payload for force-limited safety that lets them work beside people without hard guarding.

A 6-axis arm on a simple high-speed pick is slower and pricier than a delta. A SCARA can't reach around a part it needs to reorient. A cobot forced to run fast to hit takt loses the safety benefit (it has to slow near people). Match geometry to the task's reach, speed, and orientation needs.

Size the robot so the payload includes the end-of-arm tooling (EOAT) plus the part, not just the part, and check the reach envelope covers every pick and place pose with margin. Verify the allowable moment and inertia at the wrist — an offset or heavy gripper can exceed the wrist rating even when total weight is under the payload number.

The rated payload is at the mounting flange and includes everything the robot carries, so a 3 kg part on a 2 kg gripper needs a >5 kg robot with margin, not a 3 kg one. Reach must cover the *worst-case* pose (far corner, awkward orientation), which is often larger than the obvious straight-ahead distance. Moment and inertia limits catch the case where the load hangs far from the wrist center — the leverage overloads the wrist joint even at modest weight.

Sizing on part weight alone under-sizes once the gripper is added. A reach that just barely covers the layout leaves the robot straining at full extension (where it's weakest and least accurate). Ignoring wrist moment overloads the joint and shortens life or trips faults.

Use vacuum (suction) grippers for flat, smooth, non-porous, or delicate surfaces (sheet, glass, boxes, panels) and fast light picks. Use mechanical (finger) grippers for irregular, heavy, porous, or grip-around parts and where a secure positive hold matters. Consider magnetic for flat ferrous parts, and soft/adaptive grippers for varied or fragile geometry.

Gripping method follows the part's surface and geometry. Vacuum needs a sealable surface and clean, dry air but is fast, light, and gentle. Mechanical grippers hold positively around a feature regardless of surface finish or porosity, so they win on rough, heavy, or non-sealing parts — at more weight and slower actuation. The part decides: a porous casting can't be picked by vacuum; a pane of glass can't be gripped by fingers without marking.

Vacuum on a porous or textured part leaks and drops it. A hard mechanical gripper on glass or a finished surface marks or cracks it. Under-sized vacuum/grip force drops parts at speed (acceleration adds to the holding demand). Size grip force for the part *plus* the acceleration forces of the move.

Choose a cobot when humans must share the workspace, floor space is tight, tasks change often, and payload/speed are modest. Choose a traditional industrial robot with guarding for high speed, high payload, and fixed high-volume work. Either way, run a risk assessment — a cobot is not automatically safe; the whole application (gripper, part, speed, sharp edges) must be assessed.

A cobot achieves safety by limiting speed and force (power-and-force-limiting) so contact with a person stays below injury thresholds — which is exactly why it's slower and lower-payload than a guarded industrial robot. The safety lives in the *application*, not the robot: a cobot swinging a sharp tool or a heavy part at speed can still injure, so standards require assessing the full cell. Traditional robots get their speed/payload precisely by keeping people out with guarding and interlocks.

Treating a cobot as inherently safe without a risk assessment is a real hazard (and non-compliant). Forcing a cobot to run fast to hit takt erodes the safety margin. Deploying a guarded industrial robot where frequent human interaction is needed wastes the flexibility a cobot would give.

The cell can't hit the required takt time, the robot is the bottleneck, or motion is jerky and slow. Projected throughput on paper isn't matched on the floor.

Robot working near full reach (slow and weak there), inefficient path with unnecessary moves or waypoints, acceleration/deceleration limits, passing through or near singularities (where the arm slows or stalls), and waiting on feeders, vision, or grip/release settling.

1) Re-layout so picks/places sit in the robot's fast, strong mid-envelope, not at full extension. 2) Optimize the path — fewer waypoints, blended moves, avoid singularities. 3) Tune acceleration and use the robot's high-speed/anticipation features. 4) Parallelize — overlap gripper actuation, vision, and feeding with motion. 5) Check whether the bottleneck is actually upstream (feeder/vision), not the robot.

The robot misses parts, grabs them off-center, or places them out of position — dropped parts, failed insertions, and assembly jams. Repeatable in pattern or random.

Confusing repeatability with accuracy (a robot repeats well but may not be absolutely accurate), inconsistent part presentation (feeder variation) without vision, drift from thermal/wear, an uncalibrated tool center point (TCP), and grip variation.

1) Fix part presentation — consistent fixturing/feeding, or add vision to locate each part. 2) Calibrate the TCP and the robot's base/work frames accurately. 3) Distinguish the requirement: repeatability (returning to a taught point) vs accuracy (hitting a computed coordinate) and choose/calibrate accordingly. 4) Add or improve vision guidance for variable parts. 5) Address drift with warm-up and periodic recalibration.