How autonomous forklifts work: SLAM & LiDAR navigation

Every autonomous forklift runs a continuous cycle. It senses the environment with onboard sensors, thinks by working out where it is and planning how to reach its goal, and acts by driving the wheels and the mast. That loop repeats many times a second, which is what lets the truck react to a person stepping into an aisle or a pallet left out of place. The rest of this guide unpacks each stage.

The primary sense is LiDAR — laser scanning. A spinning or solid-state laser fires pulses and measures how long they take to return, producing hundreds of thousands of precise distance readings a second. Those readings form a live picture of walls, racking, pallets and people around the truck. LiDAR does double duty: it feeds navigation, and the lower, certified safety scanners feed the protective stop system.

Most platforms add other sensors to fill the gaps — wheel odometry and an inertial measurement unit to track motion, 3D cameras or depth sensors to catch low or overhanging obstacles, and fork sensors to confirm a load is correctly engaged. The sensor suite is built from established industrial components; on the platforms FlyWei integrates these are typically supplied by specialists such as SICK, RoboSense and Pepperl+Fuchs, the same kind of safety-rated laser technology used across industrial automation.

The standout idea in modern autonomous forklifts is SLAM — Simultaneous Localisation and Mapping. The challenge is circular: to know where you are you need a map, but to build a map you need to know where you are. SLAM solves both together. As the truck drives, it matches its live laser data against the map it is building, refining the map and its own position estimate at once.

The practical payoff is natural-feature navigation: the truck locates itself against the permanent features of the building — walls, columns, racking — rather than against infrastructure you have to install. There are no wires to bury, no magnetic tape to lay and no reflective markers to maintain. Commissioning is faster, and changing a route becomes a software edit rather than a floor-works project. This is the free-navigation, AMR-style approach described in our AGV vs AMR guide.

With a map and a known position, the control software plans a route to the goal — choosing a path through the aisles, respecting one-way lanes and speed limits, and computing the steering and speed commands that keep the truck on it. A separate local layer handles the immediate surroundings: if something appears in the way, the truck slows, stops or steers around it and then re-plans, rather than blindly following the original line.

Lifting is planned with the same precision. The mast height, fork position and approach are calculated so pallets are picked and placed into racking repeatably, often to within around ten millimetres, which is what makes very-narrow-aisle and high-bay storage practical without a driver.

Safety is a dedicated, certified system, not a by-product of navigation. Safety laser scanners watch configurable protective fields around the truck. As speed rises the fields extend; in tight spaces they shrink. When a person or object enters a warning field the truck slows, and entering the protective field triggers a safe stop. Movement only resumes once the path is clear.

Beyond the scanners, trucks carry emergency stops, audible and visual warnings such as light towers and blue spot lamps, and speed limits tuned to each zone. In the United Kingdom, deployment is carried out in line with PUWER (Provision and Use of Work Equipment Regulations) and LOLER (Lifting Operations and Lifting Equipment Regulations), and every site needs its own risk assessment regardless of the technology.

One autonomous forklift is useful; a coordinated fleet is transformative. A fleet manageris the software layer that assigns tasks to the right truck, schedules charging so the fleet never runs flat, manages traffic at intersections and prevents deadlocks where two trucks would otherwise block each other. It is the difference between a handful of robots and a smoothly running operation.

Increasingly this is built on open standards. VDA5050 is a widely adopted protocol that lets a single fleet manager command robots from different vendors, so operators are not locked into one supplier and can run mixed fleets. FlyWei coordinates fleets through its M4 fleet manager and RDS dispatch on VDA5050.

The robots have to fit the wider operation. The fleet manager connects upward to the systems that already run the warehouse — a Warehouse Management System (WMS), Warehouse Control System (WCS) or ERP — usually over a REST API. The WMS decides what needs to move and why; the fleet manager and the trucks work out how. An order to replenish a pick face becomes a task, the task becomes a route, and completion is reported straight back into the warehouse system.

That clean division of labour is what allows autonomous forklifts to slot into an existing operation rather than forcing a rip-and-replace. New terms here — WMS, WCS, VDA5050, SLAM — are all defined in the glossary.