Is the wrist always J4, J5, and J6?

On a typical 6-axis serial arm, yes, the three outermost axes form the wrist. On arms with a different axis count or an unusual kinematic layout, the orienting joints may be numbered or grouped differently, so check the specific model's kinematics before assuming.

Why does the robot crawl or fault when the tool barely moves?

You're probably near a wrist singularity. When J5 sits close to zero, J4 and J6 line up, and holding a smooth tool path can require those two joints to spin very fast. The controller caps that speed or throws a fault, so the tool looks like it's stalling even though the programmed move is small.

Can I just power through a wrist singularity?

No. Near the alignment there's no unique joint solution, so speeding up only makes the demanded joint velocities worse. The fix is in the path: reorient the approach or add a via-point so the wrist never passes through J5-zero with J4 and J6 aligned.

Motion

Wrist

On a 6-axis industrial arm, the wrist is the cluster of the three outermost axes (typically J4, J5, and J6) that orients the tool without moving the arm's reach.

What it is

The wrist is the compact joint group at the end of a robot arm that sets tool orientation. On a standard 6-axis arm it's axes 4, 5, and 6: J4 rolls the forearm, J5 pitches (bends) the wrist, and J6 rolls the faceplate the tool bolts to.

Think of it as division of labor. The inner axes (J1, J2, J3) position the wrist center somewhere in the work envelope; the wrist then aims the tool at whatever angle the job needs. That split is why you can hold a torch tip in one spot and pivot it through a range of angles.

How it actually works

The three wrist axes intersect (or nearly intersect) at a common point called the wrist center. Because their rotations pass through roughly the same point, they change orientation without shifting the tool's position much, which is exactly what makes a 6-axis arm able to reach a pose from several different arm configurations.

The catch shows up when J5 swings to zero, lining up the J4 and J6 rotation axes so they're parallel. Now both axes spin the tool about the same line, so the controller has two joints doing one job and no unique way to split the motion between them. This is the wrist singularity: near it, a small, steady tool-path move can demand huge, sudden J4 and J6 speeds, and the robot slows or faults to protect itself.

How it differs

Singularity · The wrist is hardware, the three orienting joints. A wrist singularity is a specific pose those joints fall into (J5 at zero, J4 and J6 aligned) where the math for splitting motion between them breaks down. One is a mechanism, the other is a configuration of that mechanism.
Reach limit · Running out of reach is the arm axes (J1-J3) hitting the edge of the work envelope. A wrist problem is about orientation: you can be well inside the envelope and still hit a wrist singularity or a J4/J6 travel limit while trying to hold an awkward tool angle.

Where you meet it in the field

FANUC LR Mate 200iD · A 6-axis tabletop arm where axes 4, 5, and 6 make up the wrist and orient the end-of-arm tooling.
Arc welding · Torch angle to the joint is set entirely by the wrist, which is why weld paths get planned to keep the wrist away from J5-zero singularities.

Common questions

Is the wrist always J4, J5, and J6?: On a typical 6-axis serial arm, yes, the three outermost axes form the wrist. On arms with a different axis count or an unusual kinematic layout, the orienting joints may be numbered or grouped differently, so check the specific model's kinematics before assuming.
Why does the robot crawl or fault when the tool barely moves?: You're probably near a wrist singularity. When J5 sits close to zero, J4 and J6 line up, and holding a smooth tool path can require those two joints to spin very fast. The controller caps that speed or throws a fault, so the tool looks like it's stalling even though the programmed move is small.
Can I just power through a wrist singularity?: No. Near the alignment there's no unique joint solution, so speeding up only makes the demanded joint velocities worse. The fix is in the path: reorient the approach or add a via-point so the wrist never passes through J5-zero with J4 and J6 aligned.