Caster Specifications for AGV Odometry and Navigation: What the Evidence Settles

Rule Out Quotes That Omit Encoder Integration or Swivel-Lead Confirmation

Two caster properties are the ones the evidence directly connects to AGV navigation, so a quote that is silent on either is not yet comparable.

Procurement guidance for AGVs is to seek casters with integrated encoders or steering drives, and to confirm swivel lead (eccentricity) with the manufacturer at the quoting stage rather than reading it off a generic catalogue 21 12 .

On the drive side, a vertical drive wheel with integrated suspension is reported to reduce the impact forces reaching the gearbox, motor and encoder, which is the mechanism behind long-term reliability on uneven floors, threshold gaps and dock levelers 33 .

That suspension finding is stated for a drive wheel, not a caster, so treat it as a design principle to ask about rather than a caster specification.

• ✓Ask whether the caster or drive unit carries an integrated encoder, or whether steering is handled by a separate drive
• ✓Request the swivel lead / eccentricity value in writing at quoting — it is rarely disclosed by generic suppliers
• ✓For the drive wheel, ask whether suspension is integrated and what impact it absorbs
• ✓Confirm the encoder is protected from the same impacts the suspension is meant to absorb

Set Swivel Offset: Turning Radius Against High-Speed Stability

Swivel offset, also called eccentricity or lead, is the horizontal distance between the swivel axis and the wheel's ground contact point, and it is what forces the wheel to trail behind the raceway centre as the cart moves 4 .

That geometry sets two behaviours at once: a smaller offset sweeps a smaller arc, while a larger offset improves high-speed stability 4 .

The turning radius follows directly from the geometry — cart turning radius equals wheelbase plus twice the swivel offset — which is why AGV and tugger specifications often use a shorter offset for aisle navigation 4 . Offset also governs low-speed behaviour.

Too little eccentricity lets the wheel hunt, an oscillation at low speed, while too much raises turning resistance and drains the battery faster 12 . For AGV casters, 15–20 mm is the range one manufacturer reports as performing best, with 20–25 mm suggested for heavier material carts that need added stability 12 .

Treat that as a reported recommendation rather than a standard, and confirm the value with the manufacturer, since it is rarely published 12 . Neither source quantifies how offset affects odometry error during a turn, so the offset decision rests on turning behaviour and stability, not on a measured positioning figure.

Wheel Material and Diameter: Rolling Resistance and Floor Protection

Rolling resistance is the push force divided by the load, and it is dominated by wheel material and diameter rather than by bearing type 3 .

The coefficient matters because it multiplies by the loaded weight to give the sustained force the AGV must overcome, so a higher coefficient is energy the drive has to supply on every metre of travel 3 . The table compares the coefficients the sources publish against what each material does for the floor.

Bearing Type: Rolling Effort Versus Load Capability

Bearing choice changes rolling effort, but it is a smaller effect than wheel material or diameter 3 . Precision ball bearings show 20–30% less rolling resistance than roller bearings, while tapered roller bearings carry heavier load at the cost of 10–15% more resistance than ball bearings 3 .

Against a plain bearing at the same load, an annular ball bearing needs roughly 40% of the effort 8 . Plain bearings are the simplest and cheapest option but offer less precision in rotation and wear faster, which is why they suit equipment that sits far more often than it rolls 8 22 .

For an AGV, the practical reading is that ball bearings give smoother, more consistent rolling at the same load, and the sources support that much.

They do not support more than that: no source in this set gives starting torque values or links bearing wear to odometry drift, so bearing selection cannot be presented as an odometry guarantee 3 8 22 .

Sizing Load Capacity: Divisor and Safety Factor

Load capacity is sized by dividing the loaded weight by the casters that actually carry it, then applying a safety factor 6 28 .

The divisor is the part buyers get wrong: on a four-caster cart that may sit on an uneven floor, the load can fall on three wheels, so dividing by four understates the real static load by a third 28 .

The safety factor is where the sources diverge — one applies 20% on top of the three-corner rule, another adds a 10–20% buffer and uprates again if loads shift during motion, and a procurement source uses 1.25 to 1.5 for dynamic load per caster 6 14 21 .

Because the factor varies with floor evenness and load dynamics, the buyer has to pick one and document the assumption rather than treat any single figure as universal 6 28 14 21 . Under-sizing is the failure mode to avoid: a caster sized below its real duty can flat-spot, brinell or chunk on the first duty cycle 6 .

No source links the load rating chosen to odometry accuracy under varying payloads, so this section protects durability, not positioning.

Match Tread and Wheel Size to the Operating Environment

Floor quality decides rolling resistance and wear, and it changes the wheel the buyer should specify. Cracked joints and debris call for larger diameters and wider treads that bridge the gaps and shed chips, which keeps resistance and wear down 14 .

Where thresholds are unavoidable, beveled ramps and a larger wheel reduce the impact 14 . Chemical and temperature exposure narrows the tread choice further.

Polyurethane is described as the only choice among common treads when oils, solvents or temperature extremes are present 5 .

The sources give no numeric temperature ranges or debris-tolerance specifications, so the buyer should state the actual exposure in the RFQ and ask the supplier to confirm the compound against it rather than assume a published range covers the application.

Standard or Custom: Up-Front Price Against Claimed Lifetime

Custom casters cost more up front and take longer to arrive, and the case for them rests on a supplier claim rather than a measured comparison.

One supplier states that a custom caster might last five times longer than a standard caster, which would lower the cost per year of use, and that standard casters may need replacement multiple times a year under continuous heavy loads 29 .

That is the supplier's position, not an independently verified figure, so it should be tested against the duty cycle rather than accepted as a budget line. Lead time is the other half of the decision.

One manufacturer publishes 2–4 weeks for standard orders and 3–5 weeks for custom, with sampling at 5–7 days standard and 15–30 days custom 30 . Those are one supplier's stated times and should be confirmed per project.

No source in this set gives tooling costs for custom encoder mounts, so that line has to be quoted directly.

Certification and Protection Claims to Verify With the Supplier

The evidence does not settle the compliance and ingress-protection questions an AGV project usually raises, so these are checks rather than specifications. No source in this set addresses ISO 3691-4 compliance, and the only IP rating present is IP65 on a drive wheel, not on a caster-mounted encoder 33 .

Ask for the documents rather than inferring them from a product family.

• ✓Ask the supplier to state whether ISO 3691-4 applies and to provide evidence — no source here addresses it
• ✓Request the IP rating of the specific caster or encoder assembly, not the drive wheel it is compared with
• ✓Ask for encoder resolution, interface type and maximum speed in writing — none are established by these sources
• ✓Confirm whether the quoted swivel lead is a nominal or a tolerance-controlled value

Where the Sources Disagree on Load Sizing

The sizing sources give different divisors and safety factors for the same decision, so the buyer should choose a factor that matches their floor and load dynamics and record it.

What the sources do not establish

• No source quantifies slip, rolling resistance, or effective rolling radius variation under load and floor conditions in relation to odometry error.
• No source provides encoder resolution (PPR), interface (quadrature, SSI, CANopen), or maximum speed for caster-mounted encoders.
• No source gives starting torque values or wear-to-odometry-consistency data for bearing types.
• No source quantifies the effect of rigid versus spring-loaded suspension on wheel-ground contact or odometry during turns.
• No source gives numeric temperature ranges or debris-tolerance specifications for environmental compatibility.
• No source quantifies how varying payload affects rolling resistance and odometry.
• No source gives tooling costs for custom encoders or mounts.
• ISO 3691-4 compliance is not addressed by any source.