Wheeled Material Racks on AMRs: Make Caster State a Mission Condition

Wheeled Material Racks on AMRs: Make Caster State a Mission Condition

In many intralogistics projects, a wheeled material rack appears to be an easy load for an AMR. It has an open underside, the robot can drive underneath, and the rack can be lifted and moved along a defined route. Yet the variation that disrupts a shift often does not come from rated payload. It comes from four casters that are not in the same condition. Different caster orientations, incompletely released brakes, debris around a wheel, or one wheel with noticeably higher rolling resistance can make the same rack behave like a different mission object at departure, through a turn, and when it is set down.

For that reason, the mission condition for a latent-lifting AMR should not stop at “rack present.” A more useful question is whether the rack is in a caster state that permits lifting, transport, and stable placement at the receiving station. Bringing that condition forward keeps random mechanical resistance from becoming something the robot must discover halfway through a route.

A caster is part of the combined load, not a rack accessory

After a rack is lifted, its casters are temporarily clear of the floor, but they still shape the combined load envelope, center of mass, and landing behaviour. Before the lift, swivel orientation affects whether the rack remains square. During entry, swivel clearance, brake hardware, and the relationship between the caster frame and the lower rack rails affect whether the robot can center itself. When the rack is lowered, the ability of all four wheels to contact the floor consistently determines whether it stands stable or shifts slightly. Treating casters as incidental hardware hides this chain of changing conditions.

A steel rack carrying totes, fixtures, or panels may be comfortably within the AMR's payload range. But if the casters create an initial skew, the robot's centerline and the rack's intended travel direction are no longer aligned. A correct route and conservative speed cannot fully remove that misalignment at a narrow aisle, a station boundary, or the moment of landing. The operational object that needs to be managed is the rack-and-caster assembly in its current state.

The same rack can have different departure resistance

Swivel casters have an important practical behaviour: under force, they rotate until their wheels trail in the direction of motion. A person pushing a rack compensates for this naturally. A latent-lifting AMR, however, depends on known geometry between its entry, lift, and departure actions. If the initial orientations are scattered, or if one caster is partly braked, binding, or obstructed, the rack can present a different initial resistance each time it was recently set down, moved a short distance, or prepared by an operator.

This does not mean every mission needs a precise rolling-resistance measurement. A more practical approach is to identify recurrent conditions: are all casters intact and touching the floor, are brakes released, do swivel joints rotate freely, can wheels be returned to an approved orientation range, and is the rack standing in a posture that allows the AMR to enter from the intended side? For a high-frequency carrier, a small number of explicit states is more valuable than one vague “movable” label.

Figure 1: A latent-lifting AMR centers under a wheeled rack before lifting

Turn caster condition into checkable mission fields

A lightweight set of mission fields can convert workshop knowledge into an operating rule. Start with carrier identity and version: racks with a similar outline should not be assumed to share the same casters or lower-frame dimensions. Then confirm the departure state, including released caster brakes, no obvious debris around the wheels, an unobstructed underside, and a load within its permitted envelope. For frequent routes or narrow aisles, a “caster orientation reset” can be a dispatch prerequisite rather than an issue discovered only after an exception stop.

Not every field requires a sophisticated sensor. A fixed departure bay can use guide edges, locating stops, or a simple station confirmation to create a repeatable rack posture before release. In manually prepared areas, caster state can be part of a replenishment, inspection, or handoff checklist. The system does not need to automate every check. It does need a clear destination for an unconfirmed state: a review position, a lower-risk route, or an operator check - not release as an ordinary transport mission.

Sequence the lift around the confirmed caster state

The latent-lifting sequence should preserve meaningful confirmation points: align with the rack centerline, enter until the lift interface is within the lower-frame support area, lift and confirm that the casters are clear and the rack posture is stable, then begin travel. The point is not to add ceremony. It is to avoid treating “the robot is under the rack” as proof that it has a transport-ready load. When caster orientation, brake condition, and load state are established at the source, this phase can verify that the actual support matches the mission condition.

The same reasoning applies at the receiving end. The drop location should give the casters a level contact surface, enough space, and any required guidance to take the load again. If the rack will be pushed by an operator or received by another device after placement, its caster orientation and brake state should not be left as an unobserved assumption.

Figure 2: The rack is on its casters while the AMR completes centered placement confirmation

Keep the exception where recovery is inexpensive

Caster issues are easiest to expose before the robot lifts the rack, or immediately after a rack is placed in a controlled station. At that point, a wheel can be reset, cleaned, released, or sent to a review bay. The impact is one carrier preparation. If the same issue appears mid-route, it can occupy an aisle, disturb adjacent missions, and require intervention.

For wheeled racks, the automation boundary is not simply whether an AMR can enter and lift. The more durable capability is that each rack waits in a predictable condition before the lift and remains in a reusable condition after placement. By making caster state a mission condition, a latent-lifting AMR can turn a seemingly simple move into a stable, repeatable logistics action.