Tugger, Shelf-Carrying, or Mobile Cobot? Matching the Right AMR Architecture to Your Warehouse Task

Choosing an autonomous mobile robot starts with one question: how does your material actually move? If it rolls on carts, a tugger AMR that pulls a convoy is likely your architecture. If it sits on shelves or totes and needs to reach a picking station, a carrying or shelving AMR fits. If the task is picking, packing, or assembly itself, you need a mobile cobot with a manipulator arm. This guide compares those three architectures on function, use cases, payload, and the operational trade-offs that decide which one belongs in your facility.

Understanding Autonomous Mobile Robot Architectures

AMRs differ from automated guided vehicles in that they navigate freely rather than following fixed paths, so the meaningful comparison is not brand versus brand but architecture versus architecture. Industry guides break the category down by what the robot does to the load. One widely referenced taxonomy of AMR types describes tuggers that pull carts like a convoy, cart-pulling robots that attach beneath carts with pin hooks, robots that carry pallets and shelves, and mobile cobots that fuse AMR movement with robotic arms.

Two navigation approaches dominate modern AMRs. Most rely on LiDAR-based simultaneous localization and mapping (SLAM), where laser scanners build a map of the facility and track the robot's position within it. Others use vision-based SLAM, which substitutes cameras for lasers. The operational distinction matters for layout flexibility: because both approaches let the robot re-map its surroundings rather than follow taped lines or embedded wire, they support facilities where aisles, racking, and workstations change over time. What differs in practice is how each sensor type copes with your environment—dust, lighting, and aisle geometry all influence reliability—which is why navigation should be evaluated against your actual floor conditions, not just a spec sheet.

A second framing decision comes before any robot is selected: identify the load first. TGW Logistics advises operators to consider the load type and load requirements before evaluating performance requirements, noting that smaller warehouse robots roughly two feet by one-and-a-half feet handle totes and cartons, while standard pallet transport vehicles measure about four feet by three-and-a-half feet, and the largest configurations carry significant payloads that vary by vendor and application. That load analysis, in TGW's words, "determines every subsequent specification decision."

Tugger AMRs: Hauling Heavy Loads

Tugger AMRs are tow tractors. They latch onto a cart—automatically or manually—and pull it like a convoy, which makes them excellent for transporting wheeled containers and dollies. The AGV Network's taxonomy highlights their core benefit: moving multiple carts in a single run rather than one at a time, and doing so without modifying the carts you already own. That last point is a genuine integration advantage, since cart-pulling alternatives that dock underneath a cart require purpose-built carts to accommodate them, even if they offer a smaller footprint and better maneuverability in narrow corridors.

Tuggers appear where bulk internal transport is constant: large manufacturing units, postal services, and hospitals moving laundry or waste, according to the AGV Network. Built In's guide adds that autonomous tuggers commonly make runs to restock items, alongside autonomous forklifts and pallet trucks in broader fleets.

The number that matters most for a tugger is payload capacity, because it directly sets trips per hour. NuminaGroup's Jen Maloney defines payload capacity as the maximum weight that can be loaded onto an AMR during fulfillment before it must be offloaded and re-inserted into the workflow, and argues that a larger-capacity AMR manages a larger workload, reducing the number of robots needed for the same output. By her figures, small AMRs handle roughly 250 pounds, while large-capacity platforms such as Numina's own AMR Batchbot reach 6,000 pounds—meaning fewer trips and more picking locations covered per run. The practical takeaway: in tugger deployments, under-specifying capacity is the most expensive mistake you can make.

AMR Architecture Comparison for Material Handling

The table below distills the three architectures by their core handling method—pulling carts, carrying or lifting load units, or manipulating items—and maps each to its evidence-supported applications and payload range. Use it as a first-pass filter: find the row whose function matches your workflow, then verify payload and load type against your actual materials.

Three primary AMR architectures compared for warehouse and manufacturing selection
AMR Architecture	Primary Function	Typical Use Cases	Key Operational Consideration	Payload Capacity (approx.)
Tugger AMR	Pulls carts in a convoy, latching automatically or manually; no cart modification needed	Large manufacturing units, postal services, hospital laundry and waste transport, restocking runs	Payload capacity directly determines trips per run and fleet size—under-sizing capacity multiplies robot count	Roughly 113 kg (250 lb) for small AMRs up to about 2,700 kg (6,000 lb) for large-capacity platforms
Carrying/Shelving AMR	Carries pallets, totes, or entire shelves/racks directly on the platform	Goods-to-person fulfillment, e-commerce hubs, sortation, warehouse material transport, pick assistance	Match load type and capacity to the material—tote-sized robots can't stretch to pallet duty	About 36 kg to 1,500 kg across surveyed manufacturer specs, depending on model
Mobile Cobot	Combines AMR mobility with a mounted robotic arm for manipulation	Flexible picking, packing, palletizing, factory assembly, delicate handling such as semiconductor wafers, postal sorting	The arm's precision and reach define the executable task; more integration complexity than transport-only AMRs	Set by the mounted arm rather than the base; no platform payload figures appear in the surveyed specs

Three primary AMR architectures compared for warehouse and manufacturing selection

AMR Architecture

Primary Function

Typical Use Cases

Key Operational Consideration

Payload Capacity (approx.)

Tugger AMR

Pulls carts in a convoy, latching automatically or manually; no cart modification needed

Large manufacturing units, postal services, hospital laundry and waste transport, restocking runs

Payload capacity directly determines trips per run and fleet size—under-sizing capacity multiplies robot count

Roughly 113 kg (250 lb) for small AMRs up to about 2,700 kg (6,000 lb) for large-capacity platforms

Carrying/Shelving AMR

Carries pallets, totes, or entire shelves/racks directly on the platform

Goods-to-person fulfillment, e-commerce hubs, sortation, warehouse material transport, pick assistance

Match load type and capacity to the material—tote-sized robots can't stretch to pallet duty

About 36 kg to 1,500 kg across surveyed manufacturer specs, depending on model

Mobile Cobot

Combines AMR mobility with a mounted robotic arm for manipulation

Flexible picking, packing, palletizing, factory assembly, delicate handling such as semiconductor wafers, postal sorting

The arm's precision and reach define the executable task; more integration complexity than transport-only AMRs

Set by the mounted arm rather than the base; no platform payload figures appear in the surveyed specs

Read the payload column with care: those figures come from manufacturer spec listings and vendor-published comparisons, and actual capacity varies by model and vendor. TGW's guidance stands—identify your load type and requirements first, then evaluate performance, because load requirements determine every subsequent specification.

Tugger, Shelf-Carrying, or Mobile Cobot? Matching the Right AMR Architecture to Your Warehouse Task

Understanding Autonomous Mobile Robot Architectures

Tugger AMRs: Hauling Heavy Loads

Carrying and Shelving AMRs: Goods-to-Person Efficiency

Mobile Cobots: Precision and Dexterity on the Move

AMR Architecture Comparison for Material Handling

Selecting the Right AMR for Your Operations

Sources

Tags

Related articles

Home Depot Buys SIMPL Automation to Speed Same-Day Fulfillment

Supervised, Shared, or Adaptive: Choosing a Human-Robot Collaboration Architecture for Manufacturing

EKF, Graph, or Visual SLAM: Which Algorithm Family Fits Your Robot's Environment

Digital Twins for Industrial Robots: Architecture, Modeling Choices, and the Path to Predictive Maintenance