Heavy-Payload AMR Suspension: Keep Every Wheel Loaded, Grounded and Useful

When Every Wheel Touches the Floor but Not Every Wheel Is Working

A heavy autonomous mobile robot can stand level, carry its rated payload and still have an unresolved ground-contact problem. One caster may carry far more than its catalog share. A powered wheel may touch the floor but carry too little normal force to produce the required traction. A diagonal floor joint may force a rigid frame to twist until one support unloads. A compliant module may protect contact yet allow the deck or sensor geometry to move beyond a functional tolerance.

This is the engineering territory of heavy payload AMR suspension. The suspension is not installed merely to make the ride feel smoother. It decides how gravity and inertial loads are shared among wheels, how vertical floor variation enters the frame, whether driven wheels remain useful, whether steering modules stay aligned and whether the vehicle can preserve its approved behavior as payload, route and component condition change.

The subject sits between several topics already covered on this site. The heavy-payload AMR load-path guide follows forces from payload to floor. The traction engineering guide explains why wheel torque is useful only when the contact patch can transmit it. The wheel-selection guide treats the wheel as a material, thermal and wear interface. This article addresses the missing mechanism between them: how the chassis keeps the right wheels loaded while the floor, motion and configuration keep changing.

The central deliverable is not a claim that the robot “has suspension.” It is a verified contact-continuity envelope: a defined set of payloads, floor inputs, maneuvers, speeds and component conditions over which critical wheel reactions stay inside their working bands without exhausting travel, destabilizing the body or damaging the frame.

Contact Continuity Is a State Map, Not a Yes-or-No Claim

A photograph cannot prove that every wheel is contributing correctly. Contact is not binary. A wheel can be touching the floor while almost unloaded, heavily overloaded, pressed against a bump stop or oscillating between separation and impact. A useful AMR wheel contact review therefore classifies the state of each wheel or wheel group instead of recording only “contact” or “no contact.”

The working band is role-specific. A free caster needs enough load to roll and swivel predictably but does not need the same minimum reaction as a driven wheel. A Mecanum wheel must transmit a resultant force through angled rollers, so losing reaction at one corner changes the achievable force set of the entire platform. A high-capacity passive wheel can remain below its bearing rating and still be unacceptable if it unloads the drive units that actually propel and stop the vehicle.

Why Four or More Supports Create an Indeterminate Load Problem

Three non-collinear support points define a plane. A perfectly rigid body placed on three ideal contacts has reactions that can be solved from vertical force and moment equilibrium. Add a fourth rigid support on a floor that is not perfectly coplanar and geometry alone no longer determines every reaction. The result depends on tire compliance, frame flexibility, joint clearances, suspension stiffness, manufacturing tolerances and the local floor elevations under the wheels.

This is why dividing total weight by wheel count is often misleading. Equal reactions require more than a centered center of gravity. They also require compatible support geometry and stiffness. A small wheel-diameter tolerance, weld distortion or floor height difference can change which corner takes additional load. With a very stiff chassis and hard wheels, the structure itself may become the equalizing spring. That can produce large diagonal twist even when every component looks generously sized under a symmetric static model.

An AGV wheel load distribution model should therefore include the actual support mechanism. A pivoting rocker is represented by its lever geometry and pivot friction. An independent spring module requires rate, preload, motion ratio, damping and travel. A hydraulic equalizer requires the connected circuit, cylinder areas, accumulator behavior and valve states. A rigid mount requires enough chassis and wheel compliance to calculate how reactions redistribute.

The purpose is not to create the most detailed model possible. It is to predict the reaction variables that control traction, wheel capacity, frame stress, docking geometry and stability under the route events that the machine will actually encounter.

Five Route Scenes Reveal More Than a Flat-Floor Load Test

Scene 1: Centered payload on a nominally level floor

This is the reference condition, not the conclusion. Measure individual wheel reactions, suspension positions and deck attitude. The sum should match the configured weight within measurement uncertainty, while left-right and front-rear differences should agree with the measured center of gravity and support geometry. An unexplained corner imbalance at this stage may indicate preload error, wheel-radius variation, assembly distortion or a floor datum problem.

Scene 2: One corner crosses a raised joint

A raised joint imposes relative wheel displacement. The event may increase reaction at the climbing corner, unload the diagonal corner and twist the frame. Speed changes the response because the input is no longer purely static. The test should record the approach, peak compression, body roll or twist, rebound and the reactions after the vehicle leaves the joint. A suspension that looks acceptable while climbing may still produce a damaging rebound impact.

Scene 3: A drive wheel crosses a local low spot

A depression can be more revealing than a bump. Passive supports may hold the frame above the low spot while the drive wheel loses normal force. Motor torque then rises without producing proportional floor force. The control system may report wheel speed, current and localization disagreement, but the root cause is mechanical contact. A spring-biased or rocker-coupled drive module is often intended to prevent this state; the boundary test must show that it actually does so for the approved depression depth and transition shape.

Scene 4: Braking and turning redistribute load

Longitudinal deceleration changes front-rear reactions. Lateral acceleration changes left-right reactions. A combined turning stop does both while asking the contact patches to transmit force. Suspension roll and pitch compliance can alter the rate and magnitude of transfer. The governing condition may be a lightly loaded inside drive wheel, an overloaded outside caster or a wheel module reaching a travel stop before the vehicle reaches its commanded deceleration.

Scene 5: Docking changes the boundary conditions

A station may guide, clamp, lift or partially support the payload. A conveyor can impose vertical and longitudinal reactions. A lift module can raise the combined center of gravity while changing the force path into the chassis. The suspension state during transfer may therefore differ from the free-travel state. If accurate docking depends on a predictable deck height, compliance and settling time become process variables rather than ride-quality preferences.

Choose a Suspension Architecture by the Reaction Behavior It Creates

The phrase AGV suspension system covers mechanisms with very different functions. Some provide only a small preload to keep drive wheels engaged. Others distribute hundreds of kilonewtons through interconnected wheel modules. Architecture selection should begin with the required reaction behavior, not with a familiar component name.

Rigidly mounted wheels

A rigid arrangement can be compact, low and mechanically simple. It may work when three principal supports define the plane, when wheel or frame compliance is intentionally used, or when the floor envelope is tightly controlled. Its weakness is sensitivity to tolerances and non-coplanar contact. Calling a wheel “rigidly mounted” does not eliminate compliance; it moves compliance into the tire, bearings, brackets, frame and floor.

Spring-biased drive modules

A vertical slide, trailing arm or pivoted drive unit can use a spring to maintain drive-wheel preload. This approach directly addresses traction starvation at a low spot. The design still needs upper and lower reaction limits. Too little preload permits slip; too much consumes wheel and bearing capacity or overloads the drive-module mount. Friction in the slide or pivot can create hysteresis, making rising and falling load paths different.

Rocker mechanisms

An AGV rocker suspension couples a drive wheel and one or more casters through a pivoting member. Lever distances establish a nominal load split while the rocker follows uneven ground. A published AGV patent illustrates how this arrangement can maintain drive-wheel traction and protect the drive unit from overload; it also shows why caster swivel direction can change the limiting geometry. The patent is an implementation example, not a universal design rule. See the original automated guided vehicle rocker-suspension disclosure.

Bogie and mechanical equalizer arrangements

An AGV bogie suspension links adjacent wheel pairs so that one wheel can rise while the assembly continues supporting the load through a pivot or equalizer. This can reduce sensitivity to isolated floor features and spread load over more contacts. The pivot and link geometry also create local forces, travel limits and maintenance points. Clearance, bearing play and lubrication condition belong in the life-cycle model because they change reaction sharing.

Independent spring-and-damper modules

Independent modules allow each wheel to follow local elevation within its travel. They offer tuning freedom but do not automatically equalize load. Spring-rate tolerance, preload setting, motion ratio and body roll stiffness determine how forces distribute. If four independent corners have different preloads, the vehicle can stand level while carrying unequal reactions.

Hydraulic or fluid-interconnected equalization

For multi-ton platforms with many wheel modules, mechanical linkages may become impractical. Fluid-connected cylinders can create defined support groups and distribute load across many axles. Wheelift describes independent fluid equalizing suspension as a core technology for its heavy-capacity AGVs, and its technical material shows how connected wheel modules can form support groups. This is a useful industrial example of AGV load equalization, not evidence that every heavy AMR requires hydraulics. See Wheelift’s heavy-capacity AGV overview and equalized-suspension concept.

The correct architecture is the simplest mechanism that keeps every role-critical reaction inside its approved band across the real configuration and route. Complexity that cannot be inspected, maintained or validated is not free performance.

Start the Calculation with Reactions, Not Spring Catalogs

For a vehicle in static equilibrium on a level surface, the vertical reactions satisfy:

These equations determine the total and the overall moment balance, but they do not uniquely solve every reaction in a multi-contact rigid structure. The suspension and compliance model supplies the missing relationships.

For a simplified rocker with one drive-wheel reaction N_D on one side of the pivot and an equivalent caster-group reaction N_C on the other, let d_D and d_C be their respective lever arms. If the pivot carries vertical load L_P and pivot friction is neglected for screening:

This is a geometry-controlled load split. It does not prove that the wheel can follow every floor feature, that the pivot remains frictionless or that dynamic reactions stay inside capacity. It gives the engineer a transparent starting point for selecting lever positions and defining what must be measured.

For each driven or steered wheel, define a role-specific reaction band:

N_{functional,min} ≤ N_wheel(t) ≤ N_approved,max

The lower bound may come from traction, braking or steering authority. The upper bound may come from wheel rating, bearing life, module structure, tire pressure, floor contact pressure or local frame capacity. Time dependence matters because a short impact peak and a sustained reaction can govern different failure modes.

Spring Rate, Damping and Travel Must Be Selected Together

A spring is not a complete suspension specification. A defensible AMR suspension design connects the load range, required static position, rate curve, preload, motion ratio, damping, friction, bump travel, rebound travel, stops and temperature behavior.

Static position and load variation

For a linear screening model, a reaction change and spring deflection are related by ΔF = kΔx. A high rate limits body movement but produces a larger reaction change for a given floor displacement. A low rate follows the floor more easily but consumes more travel and may allow excessive pitch, roll or deck-height variation. Preload establishes the initial force but does not create infinite rebound capability; once rebound travel is exhausted, contact force can still collapse.

Natural frequency is configuration-dependent

A simplified single-mode screening frequency is:

f_n = (1 / 2π) √(k / m_eff)

where k is the effective stiffness in the mode being considered and m_eff is the effective participating mass. Payload changes m_eff; tire stiffness, structural compliance and motion ratio change effective k. A frequency quoted for an empty prototype may not describe the production vehicle with its top module and maximum payload.

Damping controls the transient, not the static load split

For a simple viscous model, damping ratio can be screened as ζ = c / [2√(km)]. More damping can reduce oscillation, but excessive damping can resist rapid wheel motion and transmit higher forces across sharp inputs. Real elastomers, friction devices and dampers are nonlinear, temperature-sensitive and rate-dependent. The project should characterize the hardware over the expected temperature and velocity range instead of assuming a single constant.

Travel stops are hidden boundary-condition switches

When a suspension reaches a bump or rebound stop, its effective stiffness changes abruptly. The load path changes with it. A static model that omits stops may predict smooth reaction sharing where the real vehicle has become locally rigid. Instrumented testing should identify when stops engage and whether engagement is an accepted rare event or evidence that the operating envelope is too broad.

A Floor Profile Becomes a Time-Domain Input When the Robot Moves

The heavy-payload AMR floor-requirements guide distinguishes levelness, flatness, grade, cross-slope and local transitions. Suspension analysis consumes those route features as relative wheel displacement over time.

For a repeating spatial feature with wavelength λ crossed at speed v, the input frequency is approximately:

f_input = v / λ

This simple relationship explains why the same floor can produce different behavior at different speeds. It also explains why “slow” is not automatically safe: a long-wavelength floor wave at low speed can approach a low body mode, while a short repeating feature at higher speed may excite a wheel-module mode. Discrete steps and gaps are not sinusoidal, so their shape, edge radius, crossing angle and wheel diameter must be retained in a more complete transient model.

A 2025 physics-driven AGV simulation study provides a relevant methods example by coupling payload variation, wheel-ground interaction and suspension effects in one vehicle model. Its project model is not a universal industrial specification, but the coupling principle is directly applicable: route excitation, contact mechanics and vehicle dynamics should not be validated in separate fictional worlds. See the peer-reviewed AGV dynamics simulation case study.

Direction matters. A diagonal crossing introduces the input to wheels at different times. A Mecanum or omnidirectional platform may cross the same joint longitudinally, laterally or obliquely. The route survey should therefore preserve travel direction and speed zone, not only record a maximum joint height.

Suspension Changes the Available Traction, Brake Force and Steering Authority

A contact patch can transmit only a limited combination of longitudinal and lateral forces. For a preliminary adhesion screen, available force scales with the local normal reaction:

F_contact,max ≈ μN

The coefficient μ is condition-dependent, but the dependency on N remains fundamental. A powerful motor cannot recover force that was lost because the suspension unloaded its drive wheel. Conversely, increasing drive-wheel preload without checking wheel and bearing capacity can solve a launch problem by creating a durability problem.

During braking, the available wheel force, commanded torque and changing reaction must be evaluated together. A wheel that approaches zero reaction can lock or slip at little force. A heavily loaded wheel may require more brake torque than its drive can deliver. The site’s stopping-distance engineering guide defines the complete stop envelope; suspension provides the time-varying reactions that its force model needs.

Steered modules also require reaction. An underloaded steering wheel may rotate without producing the expected lateral vehicle response, while an overloaded module may exceed steering torque or accelerate tire scrub. Caster swivel transients can change lever arms and local drag after direction reversal. These effects should be measured during the actual maneuver, not inferred from a stationary load scale.

A Mecanum wheel suspension deserves particular attention because the platform’s commanded force depends on coordinated reactions at multiple corners and force components transmitted through angled rollers. Equal motor current does not guarantee equal floor force when wheel reactions, effective radii or roller contact states differ. The suspension must preserve useful contact without allowing body motions that compromise navigation or load-interface geometry.

The Frame and Suspension Share One Compliance Network

It is tempting to model a stiff frame and a separate compliant suspension. The real vehicle is a connected network. Tire compression, wheel bearings, module brackets, pivots, springs, top plates, crossmembers and payload fixtures all deform. The reaction distribution follows their combined stiffness.

If the suspension is too stiff for the approved floor variation, the frame may absorb diagonal displacement through torsion. That can increase weld-detail stress and move docking interfaces. If the frame is flexible and the suspension is soft, the two may create a body mode with excessive settling time. If a top module bridges multiple structural zones, it can add stiffness and reroute reactions in a way the bare-chassis model did not predict.

The chassis fatigue and deflection guide explains how repeated twist and local discontinuities become structural-life concerns. Suspension engineers should supply that analysis with wheel-reaction histories, stop-engagement events and pivot or module forces. Structural engineers should return predicted mount displacement and stiffness so the contact model does not assume an infinitely rigid frame.

This interaction is why an AMR chassis suspension cannot be approved as an isolated purchased assembly. The release configuration includes the frame, wheel modules, tires, payload structure, floor envelope and motion limits that establish its actual behavior.

Worked Screen: Three Calculations, Three Different Decisions

Consider an illustrative 2,400 kg configured vehicle, including chassis, top module and payload. The following values are not recommendations. They show how separate calculations answer separate questions.

Screen 1: rocker geometry and drive-wheel reaction

The total static weight is approximately 2,400 × 9.81 = 23,544 N. Assume two symmetric side rockers, so each pivot carries 11,772 N. On each rocker, the drive-wheel center is 0.30 m from the pivot and the equivalent caster-group reaction acts 0.20 m on the opposite side.

Using the lever relationships above:

The two drive wheels therefore carry about 9,418 N, or 40% of total weight. That percentage is created by geometry, not by motor rating.

At an illustrative adhesion coefficient of 0.25, the two drive contacts provide an approximate ceiling of 2,354 N. Suppose the vehicle accelerates at 0.40 m/s² on a 2% grade with rolling-resistance coefficient 0.02. A screening demand is approximately:

F_required ≈ ma + C_rrmg + 0.02mg ≈ 960 + 471 + 471 = 1,902 N

The nominal difference is only about 452 N before steering, disturbance or uncertainty allowances. If the surface condition reduces μ to 0.20, the approximate adhesion ceiling falls to 1,884 N, slightly below demand. The result does not prove failure or approval; it tells the team that drive-wheel reaction and low-adhesion validation are governing design variables.

Screen 2: a low spot and spring-force change

Assume a wheel module has effective vertical rate 200 kN/m. If the floor drops 8 mm beneath that wheel and body motion is temporarily neglected, the reaction reduction would screen as:

ΔN = kΔx = 200,000 × 0.008 = 1,600 N

That is large relative to the 4,709 N nominal drive reaction. The actual vehicle response will be shared by body motion, other suspensions, tires and the frame, so this is not a final prediction. It demonstrates why rate and available rebound travel must be evaluated with the complete system rather than selected from static capacity alone.

Screen 3: route wavelength and modal proximity

Assume one relevant vertical mode has effective mass 600 kg and effective stiffness 180 kN/m. Its undamped screening frequency is:

f_n ≈ (1 / 2π)√(180,000 / 600) ≈ 2.76 Hz

A repeating 0.30 m floor pattern crossed at 0.80 m/s produces an input near 2.67 Hz. The proximity does not establish resonance because damping, input shape, coupling and actual mode participation still matter. It is a reason to test that speed and route combination rather than assuming that lower speed always produces lower structural or contact response.

Together, the three screens answer different questions: geometry establishes nominal reaction, stiffness estimates sensitivity to relative displacement, and dynamics identifies route-speed combinations that may amplify response. None can replace the others.

Validation Must Reproduce Contact States, Not Just Route Completion

A serious program of mobile robot suspension testing measures wheel reactions and mechanism positions while the configured vehicle reproduces the boundary states in its intended operating envelope. Completing the route without stopping is not sufficient evidence.

Wheel-force transducers are ideal when available, but other evidence can be combined. Load cells beneath individual wheels establish static reactions. Linear displacement sensors record module travel. Strain gauges on calibrated suspension members can estimate force. IMU data captures body motion. Motor current and wheel speed identify slip or scrub. Steering angle and localization residuals reveal whether mechanical contact is degrading motion control.

Signals must be synchronized. A current spike without wheel reaction and route position cannot distinguish a joint impact from a steering transient. A minimum reaction without suspension position cannot show whether the mechanism reached a rebound stop. Time-aligned evidence converts symptoms into a causal event.

Failure Signatures Often Appear in Operations Data Before a Wheel Leaves the Floor

A suspension problem rarely announces itself as “insufficient contact continuity.” It appears as recurring system symptoms. The diagnostic task is to connect those symptoms to route position, payload state and component condition.

• Current spike plus wheel-speed divergence: possible unloading, impact or high scrub at a localized route feature.
• Direction-dependent caster chatter: swivel geometry, trail, preload or floor transition may be creating an unstable transient.
• Repeated yaw corrections on a straight path: unequal wheel reactions, rolling resistance or effective radii may be biasing propulsion.
• Docking height drift with payload: suspension sag, frame deflection or top-module compliance may be consuming interface tolerance.
• Diagonal tire wear: chronic reaction imbalance, module misalignment or chassis twist may be present.
• Cracks near wheel-module mounts: stop engagement, under-modeled impact or reaction concentration may be driving local fatigue.
• Localization disturbances at the same joint: body pitch or vibration may be moving sensors while encoders experience transient slip.

These are diagnostic hypotheses, not automatic diagnoses. The same current spike can come from contamination, a gearbox problem or a commanded acceleration. The value of fleet data is its ability to identify repeated combinations of route, load and response that deserve instrumented investigation.

Write Suspension Requirements as Measurable Boundaries

A procurement specification should not stop at “independent suspension” or “suitable for uneven floors.” Those phrases do not define what the system must accomplish. A buyer should request boundaries and evidence such as:

• approved vehicle, payload, top-module and center-of-gravity configurations;
• minimum and maximum wheel reactions by wheel role and governing load case;
• nominal ride position plus bump and rebound travel at empty and maximum load;
• spring, elastomer or accumulator tolerances across temperature and service life;
• damping or friction characterization over relevant velocity and temperature ranges;
• approved floor steps, gaps, slopes, cross-slopes and repeating surface profiles by direction and speed;
• conditions under which bump or rebound stops may engage;
• pivot, bearing, guide, seal and hose inspection requirements;
• configured-vehicle reaction, displacement and dynamic-correlation records;
• changes that trigger calculation review, adjustment or revalidation.

For a heavy duty AGV suspension, the evidence may include wheel-group pressure records, hydraulic circuit behavior and module reaction maps. For a compact AMR, it may include spring preload settings, module travel traces and drive-wheel reaction tests. The documentation scale changes; the engineering questions do not.

Standards Define Safety Responsibilities, Not a Universal Spring Rate

ISO 3691-4:2023 specifies safety requirements and means of verification for driverless industrial trucks and their systems. Its scope includes vehicles commonly described as AGVs and AMRs, and it recognizes that operating-zone condition significantly affects safe operation. It does not provide one suspension frequency, wheel-reaction split or floor-step capability for every machine. See the official ISO 3691-4:2023 record.

In the United States, the ANSI/A3 R15.08 series separates responsibilities for the industrial mobile robot, system/application integration and use. The current A3 catalog lists Part 1 as reaffirmed in 2026, Part 2 from 2023 and Part 3 for user responsibilities in 2026. These documents provide a risk and lifecycle context; they do not replace the project-specific mechanical analysis and testing needed to establish contact continuity. See the official A3 industrial mobile robot standards catalog.

Patents, supplier examples and research papers are also not generic requirements. They demonstrate possible mechanisms and methods. The engineering team must still define the configured vehicle, expected floor, payload family, motion envelope, failure criteria and verification plan.

Management of Change Must Include the Contact Model

A suspension release can become invalid without changing the nominal payload rating. The following changes can move wheel reactions or dynamic response:

• a different wheel diameter, hardness, tread profile or worn-radius limit;
• a heavier battery, relocated control cabinet or modified top module;
• a new payload footprint, CG offset, carrier or load-restraint method;
• spring replacement with a different rate, free length or preload setting;
• damper, elastomer or accumulator behavior altered by age or temperature;
• pivot wear, bearing clearance, guide friction or hydraulic leakage;
• a repaired floor, new expansion joint, route-direction change or higher speed zone;
• software changes to acceleration, jerk, braking, steering or force allocation.

Change control should ask which reaction states can move, not merely whether the new component fits. A wheel that is 4 mm smaller can change preload and effective rolling radius. A stiffer replacement elastomer can increase frame twist. A route reversal can change caster orientation before a critical floor joint. The appropriate response may range from a drawing review to a targeted reaction test or full revalidation.

A Contact-Continuity Release Review

Before production authorization, the engineering team should be able to answer the following sequence without relying on a marketing payload number:

1. Which wheels perform which functions? Identify load-bearing, driving, braking, steering and directional-force roles.
2. What reaction band keeps each role functional? Define lower and upper limits with their technical basis.
3. Which configuration and route events challenge those bands? Include empty, loaded, offset, worn and abnormal-but-credible conditions.
4. Which mechanism distributes the reactions? Record lever geometry, spring characteristics, connected circuits, stops and structural compliance.
5. Does the model predict the correct contact states? Correlate reactions, travel, body motion and frame response with hardware.
6. Does the vehicle preserve propulsion, stopping, steering and process geometry? Test system functions while the suspension is at its boundaries.
7. Can production keep the claim valid? Define inspection, adjustment, wear limits, route controls and revalidation triggers.

A “yes” at the final step should produce a controlled record: configuration, contact-state map, reaction limits, floor inputs, motion limits, analysis revisions, test data, acceptance criteria and ownership. If one link is missing, the suspension is present as hardware but not yet proven as an operating system.

Focused FAQ

Does every heavy-payload AMR need suspension?

Every vehicle needs a defined method of accommodating support tolerances and floor variation, but that method does not always require conventional springs and dampers. A three-point support, compliant tires, a pivoting rocker, structural flexibility or fluid equalization may provide the required behavior. The correct question is whether critical wheel reactions remain controlled across the approved envelope.

Why can a drive wheel slip even when the floor friction is acceptable?

The drive wheel may carry too little normal reaction because of payload position, floor height variation, suspension travel or support geometry. Available adhesion depends on both surface condition and wheel reaction. A friction measurement alone cannot prove that sufficient traction is available at every route event.

Is equal wheel load always the design goal?

No. Wheel roles and capacities differ. A design may intentionally place a defined fraction of weight on drive wheels while passive wheels carry the remainder. The goal is a controlled reaction distribution that keeps every component and function inside its approved band, not necessarily identical readings at every wheel.

Can spring preload guarantee drive-wheel contact?

Preload establishes initial force, but contact can still be lost if rebound travel is exhausted, guides bind, the floor depression exceeds the envelope or body dynamics unload the module. Preload must be assessed with rate, travel, friction, payload state and route input.

Is a softer suspension always better on uneven floors?

No. Lower stiffness can improve wheel following but increases static deflection, travel demand, body motion and settling time. It can reduce docking accuracy or allow excessive roll and pitch. Stiffness, damping, travel and process tolerances must be selected as one system.

What should be measured during a diagonal-joint test?

Measure individual or representative wheel reactions, suspension travel, stop engagement, body roll/twist, frame strain, wheel speed, motor current and IMU response. Synchronize these signals with route position and speed so the loading sequence can be reconstructed.

How does suspension affect docking accuracy?

Payload-dependent sag, body pitch or roll, local frame deflection and settling can change the height and angle of a conveyor, lift table, guide rail or fixture. Docking validation should therefore include the suspension state and external station reactions, not only navigation position.

When should suspension revalidation be triggered?

Revalidation should be considered when a change can move wheel reactions, available travel, dynamic response or functional geometry. Examples include new payload or CG limits, wheel changes, spring or damper substitutions, frame modifications, route repairs, higher speed, altered braking or steering commands and new wear limits.

Do ISO 3691-4 or ANSI/A3 R15.08 specify an AMR spring rate?

No universal spring rate is supplied for all AMRs. These standards establish safety and lifecycle responsibilities within their scopes. The manufacturer, integrator and user still need project-specific mechanical requirements, risk assessment, verification and change control for the configured vehicle and operating environment.

Conclusion: Approve the Contact State, Not the Presence of Springs

Suspension is the mechanism that negotiates a physical disagreement: the chassis wants a controlled pose, while the floor presents different elevations beneath different wheels. Heavy payload, offset center of gravity, braking, turning, docking and component wear continually change that negotiation.

A mature design does not claim success because every wheel appears to touch the floor or because a catalog lists independent suspension. It defines the role of each wheel, establishes a functional reaction band, models the mechanism that shares load, measures the governing route events and connects those reactions to traction, stopping, steering, frame life and process geometry.

The production release is therefore a contact-continuity envelope. It states which vehicle and payload configuration can cross which floor features, in which directions and at which speeds, while remaining inside reaction, travel, structural and functional limits. That is the difference between installing compliant hardware and engineering a dependable heavy-payload mobile robot.

#HeavyPayloadAMR #AGVSuspension #AMRSuspensionDesign #WheelLoadDistribution #RockerSuspension #BogieSuspension #WheelContact #LoadEqualization #MecanumSuspension #MobileRobotEngineering