How Many Carts Can an Autonomous Tugger Tow on Your Factory Route

A towing rating does not tell you how many carts will reach the line

Industry analysis | Research checked October 16, 2026

The question is timely as autonomous towing moves into additional production workflows. On October 12, 2026, Cyngn announced DriveMod EasyLaunch, a program centered on automating a repetitive indoor material movement route. On September 29, the company reported a follow-on order from an unnamed Fortune 100 heavy equipment manufacturer for another assembly-line supply application. These are company-reported commercial developments, not independent measurements of customer productivity. Cyngn EasyLaunch announcement; September 29 customer expansion announcement.

Other suppliers provide separate signals of activity. Vecna Robotics announced $31 million in funding on September 10 and described continued adoption of its autonomous forklift and tugger solutions. Shenzhen-based Ezhan announced a towing AMR on September 16 for moving carriers and material racks. Together, these announcements show recent investment, deployment and product activity. They do not establish a global market growth rate or the search volume of a particular keyword. Vecna company announcement; Ezhan product announcement.

For the buyer, the next question is practical: can the proposed vehicle use the carts already circulating through the factory? Reusing those assets can preserve established racks, packing quantities and line-side presentation. It can also carry old assumptions into a new operating model, including manual steering corrections, informal parking positions and operator decisions that the autonomous system has not been specified to reproduce.

This analysis examines autonomous tugger towing capacity through one illustrative train-selection problem. A mass calculation initially permits six carts. A staging-space calculation reduces the candidate to four. Turning, stopping, coupling and the material handoff still need to be resolved before either number becomes an approved production limit. All numerical examples below are original planning illustrations, not test results or specifications for a named product.

Define the train before comparing the tractors

An AGV tugger train consists of more than a powered vehicle and a count of trailers. The purchasing description should identify the tractor configuration, every permitted cart family, the connection arrangement, the load conditions and the routes. A six-cart train made from short tracking trailers is a different system from six long racks with swivel casters, even if both carry the same total mass.

Start by separating towing capacity vs payload. A vehicle's carried load, the gross mass of its towed carts and its available pulling force describe different physical quantities. Supplier terminology varies, so obtain the definition associated with each specification. Do not assume that every value labeled payload means cargo carried on top of the powered vehicle.

ABB's T702 product page provides a useful example of why units matter: it lists towing capacity in newtons and a separate payload value in kilograms. Those entries should not be treated as interchangeable quantities, or used to infer an unlisted deck-load rating. Ask how the manufacturer defines the supported load and the conditions attached to the force specification. ABB T702 technical data.

For a mass-based towing limit, include the tare mass of every cart, installed fixtures, containers, restraints and cargo according to the manufacturer's definition. A production planner may count only saleable parts, while the vehicle must move the complete physical train. If carts carry different products, use the actual permitted loading combinations rather than multiplying an average cart weight by the proposed count.

Keep a separate record of what the cart itself can support. The tractor's rating cannot increase a cart frame's capacity, a wheel's capacity or a connector's permissible loading. The operating limit must respect the complete assembly. For mixed carts, record where each type may appear in the train and whether the supplier has assessed that order.

Rolling behavior also deserves inspection before vehicle selection. A damaged bearing, poorly aligned caster or unsuitable wheel can change the force needed to start or turn a train without changing its measured mass. The existing guide to usable traction on the actual route explains the broader force budget. Here, the procurement task is to characterize the real carts well enough for the supplier to assess the complete duty.

Include both full and empty conditions. Return carts may have different load distributions or travel through different station arrangements. Some routes alternate loaded and empty carts, while others collect empties progressively. Those combinations belong in the operating description because a train does not remain in the same state throughout a replenishment circuit.

The worked example: six carts by mass, four by parking space

Assume a hypothetical supplier permits a maximum gross towed mass of 5,000 kg under the intended operating conditions. Each cart has a 220 kg tare mass and carries 600 kg of cargo. For this example, the tare figure includes its installed fixtures and restraints. The gross mass of one loaded cart is therefore 820 kg.

Mass-only cart count = floor(5,000 kg / 820 kg) = 6 carts

Six loaded carts weigh 4,920 kg. Seven weigh 5,740 kg and exceed the assumed mass limit. Even the six-cart result is only a preliminary screen: it leaves 80 kg below the hypothetical limit and does not establish an appropriate operational allowance for uncertain loading, additional equipment or application requirements. Actual permitted loading must be defined and controlled.

Now consider a straight waiting bay before a shared doorway. The illustrative tractor body is 2.30 m long. The distance from its rear body to the front of the first cart is 0.50 m. Each cart body is 1.60 m long, and the clear body-to-body gap between successive carts is 0.35 m. These definitions prevent drawbar length from being counted twice. For this geometry calculation, assume that loads and accessories introduce no additional front or rear overhang beyond the stated envelope.

Straight train length = 2.30 + 0.50 + (1.60 × cart count) + [0.35 × (cart count − 1)] metres

The bay offers 12.00 m of usable length. For this illustration only, the project reserves 0.75 m at each end for its planning layout. Those allowances are assumptions, not prescribed safety clearances. The final installation requires clearances and protective arrangements determined for its actual equipment and application.

The result is not that four carts are now approved. Four is the largest candidate surviving these two assumptions. The route still needs a turning assessment, stopping evidence, coupling compatibility and a workable delivery sequence. A project that purchases on the six-cart figure alone would build its capacity model around a train that cannot fit the proposed waiting location.

There are several possible responses. The plant could investigate a longer bay, shorter carts, a different replenishment batch or another route. Each changes something beyond tractor capacity. Shorter carts might reduce the amount of material carried per trip. Moving the bay could consume production space. A different route could introduce a tighter turn or a longer return cycle.

This is why increasing the tractor's towing rating may produce no benefit. An 8,000 kg tractor pulling the same carts still has to park the same physical train. The commercially useful question is which constraint prevents the required delivery service, and which feasible change resolves it at an acceptable total cost.

The first corner tests the last cart, not just the leading vehicle

A published tugger train turning radius needs a configuration and a measurement definition. It may refer to the tractor alone, a particular point on the vehicle, or an identified train arrangement. Ask which object traces the quoted radius and whether the value includes the carts, load overhang and the clearances needed for the application.

Vecna's ATG specification sheet illustrates a useful qualification: its listed 2.9 m turning radius is explicitly associated with three quadsteer carts. That is a configuration-specific statement, not a universal turning requirement for every cart train. The same sheet's minimum aisle width is a separate specification; it should not be read as proof that any train can complete any corner within that width. Vecna ATG specification sheet, 2024 Q4.

AGV trailer tracking describes how the following carts move relative to the tractor's path. Wheel arrangements, steering linkages, drawbar geometry and connection clearances influence that motion. In a turn, different points on the train can approach different obstructions. The last cart may encounter a column or stored pallet that the tractor passed comfortably.

Cyngn's official tugger FAQ identifies aisle width and cart steering behavior, alongside towing capacity, as limits on cart count. It also distinguishes the tracking behavior of quad-steer carts from other arrangements. This directly supports treating cart compatibility as a route question rather than assuming a single number applies to all carts. Cyngn DriveMod Tugger FAQ.

For the four-cart candidate, review the actual route corners in both relevant load states. Include successive turns, S-curves, narrow exits from staging and the approach to each delivery point. A train that fits a single isolated turn may still require more room when another turn follows before the carts have aligned.

The useful output is a swept-path assessment for the specified train, tied to the route drawing and followed by suitable validation. A complete train swept-path survey should distinguish fixed building geometry from space consumed by normal staging, pedestrian arrangements and other vehicles. Measure the usable route during the operating conditions the project must support.

Do not overlook reversing. A tractor's ability to drive backward does not establish that a coupled multi-cart train can reverse through the required maneuver autonomously. Confirm whether reversing is supported for that configuration, under what restrictions and with what protective arrangements. If the route is intended to remain forward-only, provide a workable turnaround and recovery path instead of relying on an undocumented reverse capability.

At pickup, the hitch must do the job the production process expects

Autonomous tugger coupling is not one universal function. A system may require a person to connect the train, release the first connection automatically, reconnect to a parked train, or handle particular carts at different stations. These capabilities create different workflows and should be written as separate requirements.

Cyngn's FAQ, for example, describes an auto-unhitch function that releases the connection to the first cart, detaching the whole train. That statement does not establish automatic delivery of individual carts at successive stops. The distinction matters if a proposed milk run assumes that the vehicle will drop one cart at each production station without assistance.

Define the desired sequence in plain operational terms: approach the correct train, establish the connection, confirm readiness, depart, arrive at the destination, secure the parked carts, release the connection and leave. For each step, identify whether the action is automatic, manual or conditional on another system. A sales description such as automated towing does not answer those details.

The mechanical interface review should include the manufacturer's permitted hitch geometry, vertical and angular movement, connector loading and cart tolerances. Floor transitions can change the relative positions of connected components. Worn or bent interfaces can also affect repeatability. A connector that works with a carefully aligned demonstration cart may not represent the condition of the customer's circulating fleet.

Connection confirmation needs an explicit meaning. A command to engage a hitch is not necessarily evidence that the required physical state has been achieved. Ask what information the control system uses, which faults it can detect, what happens when confirmation is missing and what response the application requires. The responsible integrator must determine which functions need safety-related implementation.

Parking after release is another independent requirement. Establish how the detached train remains in its intended position, how people approach it and how the next pickup is enabled. On a site with any relevant slope or surface variation, do not assume that disconnecting the tractor also secures the carts. Use the cart and vehicle manufacturers' instructions and the application's assessed controls.

The manual steps deserve honest labor accounting. If an operator must connect, inspect and release every train, include that availability in the deployment plan. Automation can still deliver value, but the business case should describe the work that actually remains. This is especially important at stations where the person expected to assist is simultaneously responsible for production equipment.

Stopping the tractor is not the same as controlling the complete train

Tugger train stopping distance must be evaluated for the configured train and intended operating conditions. The relevant system includes the powered vehicle, cart brakes where fitted, connections, load restraint and the interaction among them. A tractor-only stopping result does not establish the behavior of a long train carrying production material.

Ask how the trailing mass is controlled during deceleration and during a stop on any permitted grade. Where carts have brakes, establish their operating logic and coordination with the tractor. Where they do not, determine how the proposed system remains within its approved limits. This should be resolved by the supplier and responsible safety engineers for the actual application.

Cart arrangement can matter as well as total mass. A train with concentrated cargo at one end is not mechanically identical to a train with the same gross mass distributed across all carts. Define allowable loading patterns, load restraint and cargo overhang. Do not let operators create a new configuration simply because the total still appears below the advertised rating.

For documentation, connect the selected operating speed and load states to loaded stopping and holding evidence. Separate ordinary stopping, protective stops, emergency response and loss-of-power behavior as appropriate to the system design. This article does not prescribe a universal stop distance, deceleration value or protective-field dimension.

The validation plan should use methods approved by the supplier and the competent site team. It must not rely on a worker stepping in front of a moving train to demonstrate detection. Agree representative test configurations, instrumentation, acceptance criteria and controlled conditions before testing begins. Record the complete train state so the results can be connected to production use.

Maintenance belongs in that evidence boundary. Wheel condition, connector wear, brake adjustment and changed cargo restraints can affect the assumptions behind the released configuration. The operating instructions should identify relevant inspections, permitted replacement parts and the conditions requiring technical review before a train returns to service.

At a shared doorway, the route is occupied until the whole train clears

Tugger AGV route planning must account for the length and movement of the train, including the last cart. If a traffic controller releases a doorway or intersection when only the tractor has passed, it can give another movement permission while carts still occupy the shared space. The resource boundary must reflect the actual transport system.

Use the worked example to make the issue visible. The four-cart train is 10.25 m long in its straight configuration. At a constant 1 m/s, it would take 10.25 seconds for the rear to reach a line after the front crosses that line, ignoring articulation and other effects. This is a geometric illustration, not a proposed timer for releasing an intersection.

A real train may slow or stop while crossing. Its path may curve, and its configuration may differ from the mission record. Release logic therefore needs validated knowledge or detection of clearance, together with appropriate behavior when that information is unavailable. A fixed delay copied from the illustrative calculation would not establish that the last cart has actually cleared.

Coordinate this with the plant's intersection and shared-resource management. Door controllers, fleet software and station equipment need a consistent definition of occupied and clear. The integration should specify how cart count or train configuration is represented and what prevents an unrecognized longer train from receiving an incompatible route.

Waiting locations also need attention. A train can be short enough to cross a doorway but too long to wait beyond it without blocking the previous intersection. Review the space between shared resources, not only each resource independently. This becomes particularly important when downstream production stops and the normal flow turns into a queue.

A recovery route must accommodate the same physical train or a documented method for separating it. If the only contingency assumes manual intervention, provide access, responsibilities and approved equipment for that intervention. Production resilience depends on the ability to resolve a stopped train without creating a second obstruction elsewhere in the factory.

At the line, distinguish arriving from completing the delivery

A tugger can reach the correct coordinate while the material is still unavailable to production. The carts may require unloading, repositioning, individual identification or release by an operator. The business outcome should therefore be defined at the point where the intended material becomes available under the plant's process, not merely when the tractor reports arrival.

For a whole-train exchange, identify where the loaded train waits and where the empty train is collected. For a route serving several stations, identify which material leaves at each stop and how the remaining carts are handled. If individual carts are detached, the subsequent train configuration and route must remain within the agreed operating limits.

Use material transfer confirmation to distinguish movement completion from material acceptance. The appropriate confirmation may involve a station signal, operator action, cart identity or another agreed mechanism. The exact implementation depends on the process, but the state transition should be clear to both the robot system and the production system.

Consider a simple capacity illustration. Suppose the line needs twelve loaded carts per hour and the approved candidate carries four carts per run. That requires three completed runs per hour. If a full pickup, delivery and return cycle takes sixteen minutes, the vehicle occupies forty-eight minutes per hour with scheduled work. The remaining twelve minutes must accommodate the actual operating variability and any other required activities.

If connection delays and blocked delivery positions increase that same cycle to twenty minutes, the schedule consumes the full hour before providing any reserve. Increasing the advertised towing mass does not solve the delay if the route still accepts only four carts. The first improvement may be a better exchange position, a reliable empty-train supply or a changed station procedure.

Track return flows with the same care as loaded deliveries. Empty carts are production resources: if they accumulate at the wrong station, the next load cannot be assembled. A pilot should observe where empties wait, who prepares the next train and how unusual production sequences change the return demand. Counting successful loaded trips alone can hide a system that is gradually running out of available carriers.

Ask suppliers to demonstrate a repeatable train configuration

A useful tugger AGV acceptance test connects the intended delivery service to a defined train and route. The following proposed demonstration structure is an editorial procurement aid, not a substitute for the supplier's validation process or the application's risk assessment. The site team should establish the applicable criteria and test methods.

Do not let the demonstration silently substitute a better cart fleet for the one being purchased around. If the supplier uses new tracking trailers while the proposal assumes existing carts, record that difference. It may be a sensible upgrade, but its cost, availability and operational consequences belong in the commercial comparison.

Record results by configuration. A successful three-cart run does not prove a five-cart train, and a demonstration with one cart family does not automatically include another. Where the plant requires flexibility, ask for the permitted configuration range and the mechanism for keeping dispatch within that range. This converts flexibility from a broad promise into an operating rule.

The purchase specification should also identify ownership of cart refurbishment, hitch modifications, route changes and station integration. These items can fall between the robot supplier and the customer's maintenance or production teams. Assigning them before award is more useful than discovering during commissioning that each party assumed the other would provide them.

The final cart count should be attached to a route, a load definition and an approved workflow. That is the number operations can plan around. The tractor's towing rating remains important, but it becomes one input to a complete transport decision instead of a shortcut that promises a train the factory cannot actually use.