Automated Guided Vehicle (AGV)

AGVs follow physical or virtual paths installed in the environment rather than dynamically mapping their surroundings. Common guidance methods include:

• Magnetic tape embedded in or adhered to the floor
• Inductive wires buried in concrete that emit a radio frequency signal
• Optical lines or painted stripes tracked by onboard cameras
• Laser targets using fixed retroreflective markers for triangulation

This dependency means path changes require physical rework of the facility, making AGVs ideal for stable, high-volume routes but inflexible for dynamic environments.

AGVs operate on zone-based traffic control rather than continuous dynamic planning. Safety logic is typically binary:

• Onboard safety laser scanners detect obstacles within a defined protective field
• Detection triggers an immediate emergency stop
• The vehicle remains stationary until the obstruction is manually cleared or autonomously times out

This contrasts sharply with AMRs, which can dynamically replan around obstacles. The deterministic behavior simplifies safety certification under standards like ANSI/ITSDF B56.5 but reduces operational throughput in congested areas.

AGV fleets are coordinated by a central supervisory controller that assigns missions, manages traffic at intersections, and prevents deadlocks. Key functions include:

• Vehicle dispatching based on transport requests from ERP or WMS systems
• Intersection locking where the controller grants exclusive right-of-way to prevent collisions
• Battery management that routes vehicles to charging stations during idle periods

The central controller maintains a global state of all vehicle positions, enabling deterministic multi-vehicle coordination but creating a single point of failure risk.

AGVs are typically purpose-built for specific material-handling tasks rather than general-purpose navigation. Common form factors include:

• Unit load carriers that transport pallets or racks on a deck
• Forklift AGVs that automatically engage and lift pallets from racking
• Tuggers/tow vehicles that pull trains of non-powered carts
• Under-ride/under-cart AGVs that lift and carry mobile racks from beneath

This specialization optimizes the vehicle for its task but limits reconfigurability. Payload capacities range from 50 kg light-load units to 100+ tonne heavy haulers in steel and aerospace industries.

Most AGVs are subject to nonholonomic constraints that restrict their instantaneous motion. Common kinematic configurations include:

• Differential drive: Two independently driven wheels with casters for balance; can rotate in place
• Tricycle/Ackermann steer: Single steered drive wheel with two fixed passive wheels; cannot translate laterally
• Quad-steer: All four wheels steer independently, enabling crabbing motion for lateral docking

These constraints require path planners to generate continuous curvature trajectories that respect the vehicle's minimum turning radius, preventing the use of simple straight-line waypoint navigation.

Unlike AMRs that use SLAM for map-based localization, AGVs determine their position relative to the guidance infrastructure:

• Odometry from wheel encoders provides short-term relative motion estimates
• RFID tags embedded in the floor at known locations provide absolute position resets
• Magnetometer arrays track lateral offset from the magnetic tape centerline
• Laser triangulation computes position from angles to fixed reflectors with known coordinates

This infrastructure dependence means localization accuracy is highly repeatable along taught paths but fails completely if the vehicle deviates from the guidance system's coverage area.