Mobile Robots for Liquid Loads: Fill Level, Slosh, Container Closure, and Motion Validation

A rigid-payload calculation treats mass and center of gravity as fixed. Liquid needs an additional motion check. The same container carrying different fill volumes may remain within an AMR's payload rating while behaving very differently in motion. A passing static weight check does not establish stability or spill resistance.

This article is about the part of liquid handling that a scale cannot see: the free surface inside the container, the closure that holds it in, and the motion that the payload can actually tolerate. It is written for the buyer who has selected an AMR or mobile robot for a liquid application and now has to define what "safe to move" means.

Define the Liquid and Container Together

The container is part of the motion system, not a passive object. Its geometry and fill level shape the free surface; freeboard, closure integrity and motion determine whether liquid escapes. So the liquid and the container have to be defined together, as one payload, before any motion is planned.

Rigid containers with a tight, correctly seated lid behave very differently from open-top vessels or flexible containers. Flexible liquid bags and liners designed for the intended liquid move with their contents, which changes the dynamics again. The closure has to be verified independently of the payload: a lid that passes a static leak check may still weep under repeated acceleration and braking, because the pressure peaks during motion are not the same as the pressure at rest.

Distinguish Static Weight from Moving Liquid

When a liquid container accelerates, the free surface tilts. The liquid piles up on the trailing side and drops on the leading side. The center of gravity shifts, and it shifts differently depending on how full the container is. Headspace, fill level, vessel geometry and closure determine the available free surface and liquid response. A nearly full closed vessel may restrict free-surface motion, while a partially filled vessel can develop significant slosh; neither fill state is universally the worst case.

This is not a minor correction. A model that treats the liquid as a fixed rigid mass can estimate payload mass but does not represent free-surface slosh. Include the liquid dynamics when assessing transient motion. The literature has treated this as a control problem for decades. Feddema and colleagues studied a 230 mm diameter hemispherical water container carried by a FANUC S-800, using acceleration-profile filtering and a double-pendulum model for container tilting to suppress surface oscillations. That work is a laboratory result on a specific setup, but it establishes the principle clearly: the free surface is a state that has to be controlled, not an approximation that can be ignored.

A Chinese mainland example is the published liquid-handling patent CN111347413B from the Shenyang Institute of Automation, Chinese Academy of Sciences. Its method uses an equivalent pendulum model with vessel radius and liquid height, then adjusts the pose of an open container held by a six-axis industrial robot. This is a disclosed modeling and motion-planning method, not evidence of a commercial AMR passing a liquid-load field trial.

The practical takeaway for a buyer is that "the weight is within payload" is not the same as "the load can be moved."

For the separate mass, load-center and nominal dynamic-load checks, start with our AMR payload rating versus real load guide before qualifying the liquid motion. Slosh is the gap between them.

Compare Full, Partial, and Near-Empty Conditions

Define the complete working fill range before selecting the qualification motions.

Do not designate a worst fill level from a generic rule. For each vessel and motion profile, test low, intermediate and high working fills and identify the state that actually produces the limiting tilt, spill, pressure or stability response. A qualified operating fill range must be tied to the vessel and tested motion profile.

What a buyer can do is define fill level as a range with defined states: filled to a working maximum, typical operating fill, low fill, and the near-empty end of the range. Then validate the motion at the extreme states, not just at the nominal one. A test at 90 percent fill does not establish performance at 40 percent fill; include both states and the intervening operating range in the motion-validation plan.

The second thing to pin down is what "full" actually means for the container and the process. A fill line, a weight target, and a "visually full" check are three different criteria, and the cell should say which one governs.

Verify Container Retention and Leakage Response

Give container retention, closure integrity and leakage response their own validation line. Record how a leak is detected and how the robot or station responds.

The closure has to be verified independently of the payload. What container is used, with what closure, seated how, and checked how. If the container is a liner or bag inside a rigid outer, say so, because the two-part system fails in two-part ways. A lid that passes a static leak check may still weep under repeated acceleration and braking, so the retention check has to be run under motion, not just at rest.

Screen the liquid's hazard classification separately from motion performance. Apply the material- and jurisdiction-specific handling and packaging requirements where relevant, and define the leak response accordingly. A robot payload rating does not establish container suitability; container documentation does not establish spill-free robot motion.

Test Starts, Stops, Turns, and Handoffs

With the payload defined, the motion has to be validated against it. A steady-state rigid-load check is insufficient for transient liquid motion. Include starts, stops, turns and transfers in the validation plan.

Start the test matrix with these four motion groups, then add route-specific disturbances:

• Acceleration from rest, which tips the surface back and piles liquid at the rear
• Braking to rest, which reverses the inertial loading and must be checked against the actual deceleration profile
• Turning, especially a sustained turn or a sequence of turns, which excites lateral slosh and can build amplitude across repeated reversals
• Handoff between a robot, conveyor or station, including the actual deceleration, support changes and any impact

Turning is easy to underestimate. Yoshida and colleagues studied damped path design for a liquid container on a wheeled mobile robot across multiple turns, constraining the maximum slosh amplitude and using input shaping to keep the surface stable. That is the shape of the problem: a series of turns is not the same as one turn, because the slosh from one can feed the next.

Because the limiting fill level and the limiting motion are both setup-specific, the honest output of this stage is not a universal acceleration limit. It is a validation matrix. Test each defined fill state against each defined motion, and record what the liquid actually did. Weaver and colleagues' 2024 sensor study used an array of 14 time-of-flight sensors to reconstruct liquid motion in three dimensions and demonstrated measurements during omnidirectional mobile-robot motion. That is one research measurement approach; select instrumentation that can resolve your pass criteria. Validation with a full container at moderate speed establishes only that tested condition. It does not identify the limiting fill level or qualify the rest of the operating range.

Create the Liquid-Load Qualification Record

The deliverable from this work is a matrix. Rows are the payload and container states: container type, closure type, and fill levels from maximum to near-empty. Columns are the motion states: acceleration, braking, turning, and handoff. Each cell records the test condition, the observed behavior, and the pass criterion for that cell.

Build the matrix before specifying the cell, and the liquid-handling question stops being a guess. The fill levels that matter get tested, the closure gets checked under motion and not just at rest, the turns get run in sequence, and the hazard question is answered by classification rather than by assumption. Validate the actual states before accepting the operating envelope. A pass at one fill level cannot stand in for evidence at another.

Send the container, not just the payload weight. Share vessel drawings or dimensions, liquid type, fill range, lid or liner design, route turns, braking points, transfer method and proposed AMR model. We can use those inputs to assess your liquid-load transport application, separating basic payload compatibility from the slosh, closure and route conditions that require tests. You will have a clearer list of missing motion data and questions for the manufacturer or system integrator before a site trial is specified.