Autonomous navigation: knowing where you are in a building that keeps changing
An automated guided vehicle knows its route because somebody laid it in the floor. An autonomous mobile robot works it out, every cycle, by matching what it can see against a stored map — and the reason deployments disappoint is almost never the algorithm, it is that the building stopped looking like the map.
In one sentence
Autonomous navigation is the on-vehicle stack that lets a mobile robot establish where it is inside a facility, plan a route to where it is going, and change that route around obstacles, without wire, tape or surveyed markers installed in advance.
The stack has three parts that are usually confused with one another. Mapping builds a model of the facility once, typically by driving the robot around while it accumulates scans — the simultaneous localisation and mapping problem, which is hard because the robot must estimate its own motion and the map at the same time. Localisation is the daily job: matching a live scan against that stored map to produce a position, thousands of times a shift. Planning sits above both, computing a global route across the map and handing a local controller the job of reacting to whatever is actually in front of the vehicle.
The safety layer is deliberately not part of any of that. Navigation perception is best-effort software; the device that stops the robot when a person steps into its path is a safety-rated scanner with a certified failure rate and fixed protective fields, and it will stop the vehicle whether or not the navigation stack agrees. That separation is what makes a mobile robot certifiable, and it is also what caps its speed: the protective field has to be long enough to stop within, so a faster robot needs a longer clear corridor ahead of it.
How it works
Map decay is the characteristic failure
A warehouse rearranges pallets, stages outbound freight in aisles and moves stock daily. A map built from whatever was visible on the day it was recorded degrades against that, and a robot localising against features that have since moved drifts and eventually declares itself lost. Vendors answer it either by localising only against structure that does not move — walls, columns, ceiling features — or by updating the map continuously, which risks baking a temporarily parked trailer into the permanent layout.
Dynamic obstacles decide whether a fleet saves labour
A static map is easy. A forklift crossing an aisle, a pallet left half in a doorway and a person walking backwards are not. What matters commercially is the behaviour when the route is blocked: whether the robot waits, re-routes or gives up and asks for help. Interventions per thousand hours is the number that decides whether a deployment removed labour or relocated it to someone who walks around freeing stuck robots.
Retrofitting is a real product category
A driverless forklift does not have to be built as one. Several suppliers sell navigation as a kit that converts a standard truck from a vehicle maker, which lets an operator keep its existing service arrangements and switch a mixed fleet gradually. It also means navigation is licensed as a component rather than being the differentiator of the vehicle, which is the same unbundling that happened to motion control.
What this depends on
Technology dependencies are solved by engineering; supply dependencies are solved by building something, which takes years.
Technology
Depth and three-dimensional sensing
Obstacles that a plane scan at ankle height cannot see — overhanging forks, low pallets, a trailing cable — are only visible in depth, and missing them is how a vehicle damages stock.
A vehicle that plans only for itself will meet another vehicle that did the same; without a layer above it reserving space, two well-navigated robots deadlock in a single-width aisle.
Permitted speeds, protective field dimensions and the required behaviour on losing localisation are defined by the standards for driverless trucks; the deployment is assessed against them route by route.
What depends on this
Other pages in this map that name Autonomous navigation as something they cannot do without.
What each company supplies at this step, and — where a public figure exists — its share of this specific market — with what that share measures, the period it covers and who published it. Some rows also show the company’s own reported revenue for the segment covering this step, which is a different thing: it says how much this business matters to that company, not how much of the market it holds. Not a ranking and not a recommendation.
Supplies the Kollmorgen navigation platform that many vehicle builders license rather than write their own.
SeegridPrivate
Supplies vision-guided tow tractors and pallet trucks for North American plants.
Vecna RoboticsPrivate
Supplies autonomous forklifts and tuggers with remote human intervention built into the product.
BalyoParis
Retrofits navigation onto standard trucks from forklift makers rather than building vehicles of its own.
BlueBoticsPrivate
Licenses its navigation software to vehicle manufacturers as a component.
What would change the picture
Whether safety-rated perception improves enough to let vehicles run at full speed near people rather than slowing to a certified stopping distance.
Whether navigation consolidates onto a few licensed platforms rather than each vehicle maker maintaining its own.
Whether interventions per thousand hours fall far enough for fleets to run without on-site support.
Questions people ask about this
What is the difference between an AGV and an AMR?
An automated guided vehicle follows a route that was physically installed — wire in the floor, magnetic tape, or reflectors surveyed into fixed positions — so changing the route means changing the building. An autonomous mobile robot localises against a map of features already present, so the route is configuration rather than construction. The line is blurred in practice because most fleets are mixed and many trucks are AGVs retrofitted with map-based navigation.
Does a site have to be mapped before robots can work in it?
Yes, in almost every commercial system. The robot is driven through the facility once to build the map, and everything afterwards is localisation against it. What differs between vendors is how much the map is allowed to change afterwards, and how the system behaves when the facility no longer matches what was recorded.
Each page explains one technology in plain language, states what it depends on, and names companies by what they supply at that step. Company roles are described qualitatively and deliberately carry no market shares, revenue figures or rankings — those change faster than an explainer can, and a stale number is worse than none. Ticker links point at company pages on this site and are provided for reference only.
Nothing here is investment advice, a recommendation, or a forecast. A company named on a page about a technology is not thereby a good investment, and the chokepoints described are structural facts about supply chains rather than predictions about prices. Technology moves; where a page describes something as unresolved or in development, that was true when it was written.
Plutux no es un asesor de inversiones. Los datos de mercado y el análisis generado por IA son solo informativos y educativos, no asesoramiento de inversión. Aviso legal