移動ロボットの駆動システムの比較：2WD、4WD、4WS、および履帯式ロボット

A mobile robot’s drive system determines where it can operate, what it can carry, and how precisely it can position itself — affecting 10–25% of the total AMR BOM cost and 25–40% of energy consumption.

This guide compares the four dominant AMR drive architectures — 2WD differential, 4WD skid-steer, 4WS four-wheel steering, and tracked drive — across payload capacity, positioning accuracy, floor compatibility, energy efficiency, and total cost of ownership, so engineering and procurement teams can match chassis selection to operational requirements with data rather than marketing claims.

移動ロボットの駆動システムの比較：2WD、4WD、4WS、および履帯式ロボット

TL;DR

1. 2WD differential drive

2WD differential drive is the simplest, lowest-cost option. It works well on smooth, flat floors with light-to-medium payloads. If your facility has level epoxy or concrete floors and you’re moving totes or small pallets, start here.

2. 4WD skid-steer

4WD skid-steer adds traction and load distribution. It handles heavier payloads and slightly rougher surfaces than 2WD, but trades away some floor-friendly movement.

3. 4WS (four-wheel steering)

4WS (four-wheel steering) delivers omnidirectional maneuverability and the highest positioning precision. It is the right choice when you need ±2 mm docking accuracy, operate in narrow aisles, or handle payloads above 1,000 kg.

4. Tracked drive

Tracked drive is the outdoor and rough-terrain specialist. It crosses obstacles wheeled robots cannot, climbs steeper grades, and carries proportionally heavier loads — at the cost of speed, energy efficiency, and indoor floor wear.

No single drive system is “best.” The right one depends on your floor surface, payload requirements, space constraints, precision needs, and budget.

This guide walks through each system in detail so you can make that decision with engineering-level specificity rather than marketing generalities.

Why Your AMR Drive System Decision Determines Everything Downstream

The drive system is not one component among many. It is the architectural decision that determines how the robot moves, where it can operate, what it can carry, how precisely it positions, how much energy it consumes, and how much it costs to buy and maintain.

Change the drive system, and you change everything downstream: motor sizing, battery capacity, control architecture, safety system configuration, floor preparation requirements, and the total cost of ownership over a 5- to 7-year fleet lifecycle.

Yet in many procurement processes, the drive system gets treated as a spec-sheet checkbox rather than a strategic decision. This article is written for the engineers and operations leaders who need to make that decision with real data, not marketing claims.

What this article covers:

• The four dominant drive architectures in industrial mobile robots today are 2WD differential, 4WD skid-steer, 4WS independent steering, and tracked
• How each system works mechanically and electrically
• Performance data on payload, speed, precision, floor compatibility, and energy consumption
• A decision framework that maps your operational requirements to the right drive system
• Relevant safety standards (ISO 3691-4, ANSI/RIA R15.08) and how drive system choice impacts safety architecture

What this article does not cover:

• Mecanum and omni-wheel systems (these are specialty configurations for specific use cases; a separate article addresses them)
• Legged or hybrid locomotion (outside the scope of industrial wheeled/tracked platforms)
• Consumer or hobbyist robotics

Mobile Robot Drive Systems Comparison Table

2WD Differential Drive AMR

仕組み

A differential drive robot uses two independently driven wheels mounted on a common axis, typically at the center of the chassis. One or more passive caster wheels provide balance — two casters in a 4-wheel layout, or four casters in a 6-wheel layout for heavier loads.

Steering is accomplished entirely through the velocity difference between the left and right drive wheels. Drive both wheels forward at the same speed, and the robot moves straight. Drive the left wheel faster than the right, and the robot turns right. Drive them at equal and opposite speeds, and the robot pivots in place around its center point.

There is no steering motor, no steering linkage, no kingpin, and no servo-controlled steering angle. The drive motors do all the work, and the control system is correspondingly simple: a PID model predictive controller regulating the velocity of each wheel based on encoder feedback.

That simplicity is the architecture’s greatest strength. It’s also the source of every limitation that follows.

Key Components

1. Two drive motors

Typically brushless DC (BLDC) with integrated planetary or harmonic gearboxes. Common power range: 200–750 W per motor for warehouse-class AMRs.

2. Incremental encoders

One per drive motor, typically 1,024–4,096 PPR (pulses per revolution). These provide the velocity and position feedback for the control loop. Differential drive systems lack absolute steering encoders because there is no steering axis to measure.

3. Passive casters

Polyurethane-tread casters, usually 75–150 mm in diameter, with double-bearing swivel heads. These are wear items replaced annually in 24/7 operations.

4. Motor drivers/controllers

CANopen- or EtherCAT-networked drives, often with integrated STO (Safe Torque Off) for functional safety compliance.

5. Suspension (optional)

Spring-loaded caster mounts or a central pivot to maintain ground contact on slightly uneven floors. Many entry-level differential-drive robots lack suspension and rely on floor flatness.

Performance Characteristics

1. Payload

In standard industrial configurations, 2WD differential robots handle payloads up to 300–500 kg. The limiting factor is not motor torque — you can always spec larger motors. The real constraint is caster load capacity and floor pressure. Each caster carries roughly half the payload plus a fraction of the robot’s own weight. With a 4-caster configuration using 75 mm polyurethane casters rated at 150 kg each, the practical payload ceiling is around 500 kg. We’ve seen integrators push this to 600 kg with 100 mm casters, but at that point, the floor loading starts to cause maintenance headaches.

Above 500 kg, the casters begin to wear rapidly, floor marking becomes a problem, and the robot’s stopping distance under load extends beyond what safety standards permit without upgrading to larger, more expensive casters and motors. At that point, 4WD or 4WS becomes the more economical choice.

2. Speed

1.5–2.0 m/s (5.4–7.2 km/h) is the typical operating range for warehouse-class differential-drive AMRs. This is fast enough for most intralogistics tasks — a robot moving at 1.8 m/s covers a 100-meter warehouse aisle in under a minute. Faster speeds are possible but trigger additional safety sensor range requirements under ISO 3691-4.

3. Positioning accuracy

 Manufacturers quote ±10 mm for differential-drive robots under real-world conditions. That number assumes a well-maintained floor, solid tires, and regular encoder calibration. Throw in dust, moisture, or unevenness, and accuracy degrades quickly — wheel slip introduces odometry errors that the encoder alone cannot catch.

This is the fundamental limit of differential drive precision. The system infers steering angle and position indirectly from wheel speed. When a wheel slips, the encoder reports movement that never happened. The control system has no independent way to know it was lied to. IMUs and SLAM-based localization compensate at the navigation level — but dead-reckoning precision at the drive level stays limited no matter how good your navigation stack is.

4. Floor requirements

Smooth, flat, level, and clean. Differential drive robots need continuous ground contact from both drive wheels. A dip, ridge, or debris that lifts one drive wheel even a few millimeters causes the robot to veer off course because the lifted wheel spins freely while the grounded wheel drives the robot in an arc. Epoxy-coated concrete floors in good condition are ideal. Polished concrete with a coefficient of friction below 0.5 can cause slip. Floor flatness should meet or exceed FF35/FL25 per ACI 117 specifications.

5. Energy efficiency

Differential drive is the most energy-efficient wheeled configuration. With only two drive motors and no steering actuators consuming power, the parasitic losses are the lowest of any drive architecture. A typical 200 kg payload differential-drive AMR consumes 400–800 W in steady-state travel on a flat floor, translating to 6–8 hours of operation from a 48V/60Ah LiFePO4 battery.

6. Cost

The drive system — two motor+gearbox+encoder+driver assemblies, two drive wheels, and a set of casters — runs $2,000–$5,000 at the BOM level for industrial-grade components. That’s your baseline. Every other configuration is measured against this number.

When to Choose 2WD Differential Drive

Choose differential drive when:

• Your payloads are under 500 kg
• Your floors are smooth, flat, and in good condition
• Your routes are relatively simple (point-to-point, no tight maneuvering)
• You are deploying at scale, and per-unit cost matters
• Your docking tolerance is ±10 mm or looser
• You want the simplest maintenance profile possible

Do not choose differential drive when:

• Your floors have slope changes, expansion joints, or unevenness
• You need lateral (crab) movement capability
• You need docking accuracy tighter than ±5 mm
• Your payload exceeds 500 kg
• You operate in wet, dusty, or debris-prone environments where wheel slip is likely

4WD Skid-Steer Mobile Robot Drive

仕組み

A 4WD skid-steer robot uses four driven wheels arranged in two pairs, one on each side. All wheels on each side are mechanically linked or electronically synchronized to rotate at the same speed. Steering is achieved through a speed differential between the left and right sides — the same principle as a 2WD differential, but with four driven wheels instead of two.

The key advantage over 2WD is traction. With four contact patches instead of two, the robot can transmit more motor torque to the floor before wheel slip occurs. This directly translates to higher payload capacity, better hill-climbing ability, and more reliable movement on floors with moderate contamination (dust, light moisture).

The trade-off is the skidding. When a 4WD robot turns, the outside wheels slide laterally across the floor because they’re locked in orientation. On polyurethane wheels on epoxy, you’ll hear a distinctive squeak — and you’ll see accelerated tire wear, especially on the front and rear wheels that travel the most lateral distance during turns. If you’ve ever watched a skid-steer loader work on concrete, you know exactly what this sounds like.

Key Components

1. Four drive motors

One per wheel, typically 200–750 W each for warehouse-class robots. BLDC with planetary gearboxes remains the standard. Some designs use two larger motors to drive each side through a mechanical linkage (belt or chain), which reduces the motor count but adds mechanical complexity.

2. Motor synchronization

Electronic synchronization via CANopen or EtherCAT is the modern approach. Older or lower-cost designs use mechanical coupling (a single motor per side with a drive shaft and differential), but this eliminates independent wheel control.

3. Suspension

More critical in 4WD than in 2WD. If one wheel loses ground contact on an uneven surface, the robot loses traction on that side in proportion to the loss of contact. Rocker-bogie or independent swing-arm suspension is common on higher-end 4WD robots. Spring preload must be carefully tuned — too soft and the robot sways under load; too stiff and the suspension doesn’t articulate.

4. Tires

Larger diameter than 2WD equivalents (typically 200–400 mm vs. 125–250 mm) to accommodate the suspension travel and distribute the higher loads. Polyurethane remains standard; pneumatic tires appear on outdoor-capable 4WD robots.

Performance Characteristics

1. Payload

4WD skid-steer robots routinely handle 200–2,000 kg. The lower end (200–500 kg) is for compact warehouse robots; the upper end (1,000–2,000 kg) is for pallet-moving AMRs and autonomous forklifts. With four contact patches distributing the load, the floor pressure per wheel is lower than that of a 2WD robot carrying the same payload, thereby reducing floor wear and caster-related failure modes.

2. Speed

1.5–2.5 m/s. Faster than 2WD in practice because the additional traction allows higher acceleration and deceleration without wheel slip. Top speed is typically limited by the safety sensor’s range, not the motor’s capability.

3. Positioning accuracy

±5–10 mm. Better than 2WD because reduced wheel slip makes encoder-based odometry more reliable. However, the skid-steer turning mechanism introduces its own error source — during a turn, the lateral tire scrub is unpredictable and varies with floor condition, tire wear, and load distribution.

4. Floor requirements

Flatter and more robust than 2WD requirements. Skid-steering exerts lateral forces on the floor surface during every turn. Epoxy coatings must be properly bonded and have sufficient thickness to resist delamination. Concrete floors should be sealed to prevent dust generation from tire abrasion. The floor surface will show visible wear patterns along the robot’s routes within weeks of deployment — this is normal and should be budgeted for in facility maintenance.

5. Energy efficiency

Lower than 2WD due to four motors and the energy lost to tire scrub during turns. Expect 15–25% higher energy consumption per meter traveled compared to a 2WD robot of similar weight. In a typical warehouse AMR application, a 4WD robot consumes 600–1,000 W in steady-state travel.

6. Cost

The drive system costs 1.5–2× as much as the 2WD equivalent. Four motor+gearbox+encoder assemblies, four drivers, four wheels, and a suspension system add up. The suspension is the single largest cost adder — a proper independent suspension with machined swing arms and spring-damper units can cost as much as the motors it supports.

When to Choose 4WD Skid-Steer

Choose 4WD skid-steer when:

• Your payloads are 500–2,000 kg
• Your floors are generally flat but may have minor imperfections
• You need better traction than 2WD for acceleration, deceleration, or mild inclines
• Your budget supports a 1.5–2× cost premium over 2WD
• You accept some floor surface wear from skid-steering

4WS Four-Wheel Steering AGV/AMR

仕組み

A 4WS robot uses four independently steerable drive wheels — one at each corner of the chassis. Each wheel assembly (sometimes called a “steering wheel drive” or “steer drive”) integrates two motors: a traction motor that drives the wheel forward and backward, and a steering motor that rotates the entire wheel assembly around a vertical axis.

This dual-motor-per-wheel architecture gives the control system independent authority over both the speed and the steering angle of every wheel. The result is true omnidirectional mobility: the robot can move forward, backward, sideways (crab mode), diagonally, or rotate in place, and it can switch between these modes without stopping.

The precision advantage comes from the absolute encoder on each steering axis. Unlike differential drive, which infers steering angle from wheel speed (an indirect measurement vulnerable to slip), 4WS measures the actual steering angle directly at each wheel. A typical steering encoder has a resolution of ±0.1°, which translates to sub-millimeter lateral control at the chassis level.

Key Components (Per Wheel)

1. Traction motor

BLDC with planetary gearbox, 200 W–2 kW depending on payload class. Typically runs at 24–48 VDC with integrated holding brake.

2. Steering motor

Smaller than the traction motor (100–500 W), with a high-ratio worm or harmonic gearbox to hold the steering angle against road forces. The gearbox must be back-drivable enough for the steering encoder to detect external disturbances, but stiff enough to hold position under load.

3. Absolute encoder (steering axis)

This is the component that defines the architecture’s precision advantage. Typically a multi-turn absolute encoder with 17–19 bit resolution, providing true angle measurement even after power cycles. No homing routine needed at startup.

4. Incremental encoder (traction axis)

Standard quadrature encoder, 1,024–4,096 PPR.

5. Steering bearing

A large-diameter slewing ring or crossed-roller bearing that supports the entire vertical load of that corner of the robot plus payload while allowing smooth rotation. This is a high-cost, high-precision component.

6. Driver/controller

A dual-axis servo drive that closes the loop on both steering angle and traction velocity simultaneously. EtherCAT with FSoE (FailSafe over EtherCAT) for functional safety integration.

Performance Characteristics

1. Payload

4WS is the architecture for heavy payloads. Single-steer-drive units from manufacturers such as SEW-Eurodrive and CFA offer load capacities ranging from 500 kg to over 6,500 kg per wheel assembly. A four-wheel 4WS robot using appropriately sized steer drives can handle payloads that would be impractical with any other configuration.

This is why 4WS is the dominant architecture for autonomous forklifts, heavy pallet transporters, and outdoor logistics platforms. When you need to move 2,000 kg of material and dock it with ±2 mm repeatability at a production line, 4WS is the only wheeled architecture that can deliver both requirements simultaneously.

2. Speed

1.0–2.0 m/s. The top speed is typically limited by safety considerations and the steering system’s ability to maintain angle control under dynamic loads, not by motor power. At higher speeds, the steering motors must work harder to hold the wheel angle against cornering forces, and the control loop bandwidth becomes the limiting factor.

3. Positioning accuracy

±2–5 mm at the docking point, with repeatability as tight as ±2 mm under controlled conditions. This is the architecture’s defining advantage. The combination of direct steering angle measurement (absolute encoders), independent wheel control, and the ability to correct position using lateral movement without reorienting the chassis enables precision that differential and skid-steer systems cannot match.

This matters most in applications such as semiconductor wafer handling, precision assembly-line feeding, and hospital specimen delivery, where a 10 mm positioning error can result in a failed pickup or a collision.

4. Maneuverability: Omnidirectional. The four standard modes are:

• Ackermann mode: Front wheels steer, rear wheels fixed — like a car. Efficient for long, straight travel.
• Crab mode: All four wheels steer to the same angle. The robot moves diagonally or laterally while maintaining chassis orientation. Critical for narrow-aisle docking.
• Zero-radius rotation: Wheels steer tangentially to a circle centered on the robot. The robot rotates in place without translation—less floor-wearing than skid-steer rotation.
• Independent mode: Each wheel sis teered to its own angle. Used for specialized maneuvers, such as rotating around one corner.

5. Floor requirements

Less demanding than differential drive for flatness because the independent suspension on each corner maintains ground contact across moderate surface variation. However, the floor must provide consistent friction — a slippery patch under one wheel during a precision maneuver can cause that wheel to lose steering authority, and the robot’s localization will detect the error and stop, causing the maneuver to fail.

6. Energy efficiency

The lowest of the wheeled configurations. Eight motors (four traction + four steering) consume power even when the robot is stationary, because the steering motors must hold position against external forces. In practice, 4WS robots consume 25–40% more energy per meter than equivalent 2WD robots. This translates to larger batteries, longer charge times, or reduced operating hours for the same battery capacity.

7. Cost

The drive system is 3–5× as expensive as a 2WD differential. Do the math: four dual-motor steer-drive units with absolute encoders, four dual-axis servo drives, four slewing bearings, and the wiring harnesses. One industrial-grade steer-drive unit — motor, gearbox, encoder, bearing, no driver — costs $1,500–$4,000 depending on load class. Multiply by four. Add four dual-axis drivers at $1,000–$2,000 apiece. You’re at $10,000–$24,000 per robot in drive system cost alone — before the chassis, battery, sensors, or control system. That’s the price of precision.

This cost makes sense when precision and payload requirements demand it. It does not make sense when a 2WD differential robot would meet the same operational requirements at a quarter of the drive system cost.

When to Choose 4WS (Four-Wheel Steering)

Choose 4WS when:

• You need docking accuracy of ±5 mm or better
• Your facility has narrow aisles that require lateral (crab) movement
• Your payload exceeds 1,000 kg
• You need the robot to change direction without reorienting its chassis (attached conveyors, sensitive payloads)
• Your floors are uneven enough that a 2WD differential would lose traction
• Your application requires the robot to maneuver in multiple directions in confined spaces

Do not choose 4WS when:

• A 2WD or 4WD system meets your precision and payload requirements at a lower cost
• Your team is not prepared for the maintenance complexity (four dual-motor assemblies, eight encoders to calibrate, more failure modes)
• Your energy budget is constrained (4WS carries a 25–40% energy penalty over 2WD)

Tracked Mobile Robots

仕組み

A tracked drive system replaces wheels with continuous tracks — reinforced rubber or steel belts that run around a set of drive sprockets, idler wheels, and support rollers on each side of the robot. Each track is driven independently by a motor through a drive sprocket, and steering is by skid-steer: vary the speed of the left track relative to the right.

The defining mechanical characteristic is the large contact patch. Where a wheeled robot distributes its weight across four small contact patches (the tire footprints), a tracked robot distributes the same weight across the entire length and width of both tracks. This dramatically reduces ground pressure, which is why tracked vehicles can traverse soft surfaces that wheeled vehicles sink into — sand, mud, snow, loose gravel, and soft soil.

The trade-offs are mechanical complexity, lower speed, higher energy consumption, and the risk of damage to indoor floor surfaces.

Key Components

1. Drive sprockets

Steel or aluminum, with teeth that engage the track’s drive lugs. The motor directly drives the sprocket through a planetary gearbox. Larger sprockets reduce track tension and wear but increase the robot’s overall width.

2. Idler wheels

Unpowered wheels at the front and rear that maintain track tension and guide the track around its circuit. Typically mounted on tensioning mechanisms (threaded adjusters or spring-loaded) to set the correct track tension.

3. Road wheels/support rollers

Smaller wheels between the idlers that support the robot’s weight along the bottom run of the track. Number and spacing are critical design parameters — more road wheels distribute ground pressure more evenly but add weight and rolling resistance.

4. Tracks

Reinforced rubber with internal steel or Kevlar cords is standard for industrial robots. Steel tracks with rubber pads are used in extreme-duty applications (e.g., construction and mining). Track width and tread pattern are selected for the target terrain.

5. Motors

Higher torque than equivalent wheeled robots because the track drive is mechanically less efficient. A tracked robot carrying the same payload as a wheeled robot needs roughly 25–50% more motor torque to overcome track friction.

6. Suspension

Road wheels are typically mounted on a bogie or torsion-bar suspension to maintain track contact with irregular terrain. This is not optional — a tracked robot without suspension loses track tension and throws tracks on rough terrain.

Performance Characteristics

1. Payload-to-weight ratio

Tracked robots carry a proportionally greater payload relative to their weight than wheeled robots. A 100-kg tracked robot can carry a 100-kg payload (1:1 ratio). A comparable wheeled robot carries 25–33% of its own weight. This makes tracked drive attractive for applications where the robot itself must be small and light, but the payload is substantial — reconnaissance, disaster response, agricultural sampling.

2. Terrain capability

Tracked robots climb steeper grades (up to 34° tested, vs. ~31° for wheeled), cross obstacles 60–100% taller, and maintain better path tracking on slopes above 15°. On soft surfaces — sand, mud, snow, loose soil — tracked robots move where wheeled robots sink and stall. This is the architecture’s defining advantage and the reason it dominates outdoor, off-road, and all-terrain applications.

3. Speed

0.5–1.5 m/s for industrial tracked robots. This is slower than wheeled equivalents (which typically achieve 1.5–2.5 m/s). The speed limitation has two causes: track friction increases with speed, and tracked vehicles are harder to stop because the large ground-contact area that provides flotation also results in greater momentum transfer to the ground during braking.

4. Positioning accuracy

±10–20 mm, the lowest of the four architectures. Track slip is inherent to skid-steering — the tracks must slide laterally across the ground during turns, and the amount of slip varies with surface type, surface moisture, track tension, and track wear. Odometry from motor encoders is correspondingly unreliable. GPS, visual SLAM, or external localization infrastructure (e.g., reflectors or magnetic tape) is required for precise positioning on tracked robots.

5. Indoor floor compatibility

Poor. Rubber tracks on epoxy or polished concrete leave visible marks. Steel-reinforced tracks can gouge coated floors. The skid-steer action scrubs the floor surface. Tracked robots are generally not recommended for finished indoor floors. If a tracked robot must operate indoors (e.g., a construction robot moving between indoor and outdoor areas), floor protection measures are required, and the facility should budget for regular floor refinishing.

6. Energy efficiency

The lowest of all drive architectures. Track friction, the large number of rotating elements (sprockets, idlers, road wheels), and the continuous deformation of the track as it cycles around its circuit all consume energy. Tracked robots consume 25–30% more power than wheeled robots on the same surface and 50% or more on rough terrain. A tracked robot that operates for 4 hours on a battery may have a wheeled equivalent that runs for 6–8 hours.

7. Cost

The drive system is 1.5–2.5× as expensive as a 2WD differential. Tracks are consumable items — a set of industrial rubber tracks costs $500–$2,000 and lasts 1,000–3,000 operating hours, depending on the terrain’s abrasiveness. Track replacement is a significant maintenance line item that does not exist for wheeled robots.

When to Choose Tracked Drive

Choose a tracked drive when:

• Your primary operating surface is outdoor, unpaved, soft, or uneven
• You need to climb slopes above 20°
• You need to cross obstacles taller than 80 mm
• Your payload-to-robot-weight ratio needs to be high (0.5:1 or greater)
• Indoor floor wear is not a concern (or you only operate indoors for short transitions)

Do not choose a tracked drive when:

• Your primary operating surface is a finished indoor floor
• You need positioning accuracy better than ±10 mm
• You need travel speeds above 1.5 m/s
• Battery life and energy efficiency are critical operational constraints
• You are deploying at scale and want to avoid track replacement as a recurring maintenance cost

Mobile Robot Drive Systems Head-to-Head Comparison by Decision Criteria

ペイロード容量

2WD Differential:     ████████░░░░░░░░░░   Up to 500 kg
4WD Skid-Steer:       ██████████████░░░░   Up to 2,000 kg
4WS Steering:         ██████████████████   Up to 6,500+ kg
Tracked:              ████████████░░░░░░   Up to 2,000+ kg
                      (but excellent payload-to-weight ratio)

Positioning Accuracy (lower is better)

The difference between ±10 mm and ±2 mm may sound small. Still, in a production environment, it is the difference between a robot that docks reliably 99.5% of the time and one that requires manual intervention several times per shift.

Floor Surface Compatibility

Cost Structure Over 5-Year Lifecycle

These are order-of-magnitude estimates for a single robot operating one shift per day. A 24/7 operation running 3 shifts multiplies consumable and maintenance costs by approximately 2.5–3×. A fleet of 10 robots multiplies by 10, but with some economies of scale on spare parts inventory and technician time.

How to Choose the Right AMR Drive System?

The drive system decision reduces to five questions. Answer them in order.

Question 1: Where will the robot operate?

• Smooth indoor floors (epoxy, polished concrete): 2WD, 4WD, or 4WS are all viable. 2WD is the cost-optimal starting point.
• Rough indoor floors (unsealed concrete, cracks, joints): 4WD or 4WS. 2WD will lose traction and drift.
• Outdoor, unpaved, soft, or sloped surfaces: Tracked. Wheeled configurations will not perform reliably.
• Mixed indoor/outdoor: 4WS (with pneumatic tires) or tracked, depending on the indoor floor finish tolerance.

Question 2: What is your maximum payload?

• < 500 kg: 2WD differential is sufficient and most cost-effective.
• 500–1,000 kg: 4WD skid-steer or 4WS. 2WD is at its limit.
• 1,000–2,000 kg: 4WS is the standard choice. 4WD is possible with appropriate motor sizing.
• > 2,000 kg: 4WS is the only practical wheeled architecture. Tracked is an option only if the surface demands it.

Question 3: What positioning accuracy do you need?

• ±10 mm or looser: 2WD differential is adequate. (Typical for goods-to-person picking, zone-to-zone transport.)
• ±5–10 mm: 4WD or 4WS. (Typical for conveyor interface, pallet pickup.)
• ±2–5 mm: 4WS is the only architecture that reliably achieves this. (Typical for production line docking, precision assembly, semiconductor handling.)

Question 4: How constrained is your space?

• Wide aisles (>3 m), simple routes: 2WD differential. In-place rotation via differential steering is sufficient.
• Narrow aisles (2–3 m): 4WD or 4WS. 4WS with crab mode provides the most flexibility.
• Very narrow aisles (<2 m), tight maneuvering: 4WS. Crab movement is the only way to dock laterally in a confined space.
• Outdoor, open terrain: Tracked or 4WD. Space constraints are less relevant; terrain capability dominates.

Question 5: What is your budget envelope?

• Minimize per-unit cost, deploying at scale: 2WD differential.
• Moderate premium for more capability: 4WD skid-steer.
• Performance-driven, per-unit cost secondary: 4WS.
• Terrain capability at any indoor-floor cost: Tracked.

Decision Matrix

Safety Standards and AMR Drive System Design Implications

The drive system is directly implicated in functional safety. Safety standards do not specify which drive architecture to use. Still, the architecture determines how safety functions are implemented — and how much they cost.

ISO 3691-4: Driverless Industrial Trucks

ISO 3691-4:2023 governs the safety of driverless industrial trucks (automated guided vehicles and autonomous mobile robots in industrial environments). Key requirements that interact with the drive system design:

1. Safe Torque Off (STO)

Every drive motor must support STO — the ability to turn off torque output while maintaining the mechanical connection electrically. This is typically implemented via the motor driver (STO inputs on the servo drive). In a 2WD robot, you need two STO channels. In a 4WS robot, you need eight (four traction + four steering).

2. Braking performance

The robot must stop within a specified distance under maximum load and maximum speed. The braking system (electromagnetic holding brakes on the motors, plus any mechanical brakes) must be sized for the worst case. A 4WS robot moving 2,000 kg at 2.0 m/s needs substantially more braking torque than a 2WD robot moving 250 kg at 1.5 m/s.

3. Stability

The robot must not tip under specified slope, speed, and load conditions. The drive system’s center of gravity, wheelbase width, and suspension geometry all factor into the stability calculation. Tracked robots have an inherent stability advantage due to their wide footprint and low center of gravity.

4. Speed monitoring

The safety controller must independently monitor the robot’s speed and trigger a protective stop if it exceeds the configured limit. This requires redundant speed sensors — typically the motor encoders plus an independent sensor (radar, second encoder, or external measurement system).

ANSI/RIA R15.08: Industrial Mobile Robot Safety

ANSI/RIA R15.08 is the U.S. standard for industrial mobile robot safety. It classifies robots into Type A (autonomous mobile robots without a manipulator), Type B (AMRs with a manipulator), and Type C (mobile manipulators performing tasks at multiple locations). Drive system implications:

1. Type A applications (transport-only AMRs)

Any drive architecture can comply. The safety burden is primarily on the navigation and obstacle-detection systems.

2. Type B/C applications (AMRs with conveyors, robot arms)

Docking precision becomes a safety issue — if the robot does not dock accurately, the manipulator or conveyor interface may fail in a way that creates a hazard. This pushes toward 4WS when docking tolerance is tight.

Functional Safety Cost by Architecture

The number of motors directly determines the number of STO channels, safe-speed-monitored axes, and safety-rated encoder interfaces. In rough terms:

• 2WD: 2 STO channels, 2 safe-speed axes. Lowest safety BOM cost.
• 4WD: 4 STO channels, 4 safe-speed axes. Moderate safety BOM cost.
• 4WS: 8 STO channels (4 traction + 4 steering), 8 safe-speed axes. Highest safety BOM cost. Safety-rated absolute encoders on the steering axes add $200–$500 per wheel.
• Tracked: 2 STO channels, 2 safe-speed axes. Similar to 2WD, but the mechanical braking system may need to be larger due to the robot’s all-terrain momentum.

These costs are not trivial. For a 4WS robot, the functional safety hardware alone — safety-rated drives, safety PLC, safety encoders, safety sensors — can exceed the BOM cost of a complete 2WD drive system. We’ve seen integrators spec 4WS for applications that a 2WD robot could handle with a few floor repairs and a wider aisle. That’s not engineering — that’s spending money you didn’t need to spend. When the application genuinely demands 4WS performance, the safety BOM is a necessary line item. When it doesn’t, it’s an argument for simplicity.

Conclusion: Match the Mobile Robot Drive System to Your Operation

After 3,000+ words comparing four drive architectures across a dozen criteria, the conclusion is not that one system is best. The conclusion is that the drive system decision is a matching exercise between your operational requirements and each architecture’s strengths and weaknesses.

Most warehouse automation projects should start by evaluating a 2WD differential drive. It is the simplest, cheapest, and most energy-efficient option, and it works for most indoor logistics applications. Only move up the complexity ladder when a specific operational requirement demands it.

4WD is the logical first step up: more payload, more traction, and a slightly higher cost. 4WS is the precision-and-payload leader, and its cost premium is justified when ±2 mm docking or omnidirectional movement in tight spaces is a hard requirement. Tracked drive is the outdoor specialist — essential for rough terrain, inappropriate for finished indoor floors.

The most expensive mistake isn’t choosing the wrong architecture. It’s choosing an over-specified one for an application that didn’t need it. A fleet of 20 AMRs with 4WS drive systems in a warehouse with wide aisles, smooth floors, and ±10 mm docking tolerance — that’s $300,000–$500,000 that should have gone into a second shift of 2WD robots, or better end effectors, or a mezzanine expansion. Match the drive system to the requirement. Not to the spec sheet.

Ready to specify a drive system for your mobile robot application?

Contact Fdata’s engineering team to discuss your payload, floor, precision, and budget requirements. We build custom mobile robot chassis with 2WD, 4WD, 4WS, and tracked configurations for B2B industrial applications worldwide.

よくあるご質問