Ocado Smart Platform Warehouse Robotics
by Nick Clark | Published April 25, 2026
Ocado Group licenses the dominant online-grocery fulfillment platform to retailers worldwide, operating dense robotic warehouses with thousands of bots picking customer orders at supermarket scale. The architectural element it does not provide — a credentialed lane-authority substrate that lets robots, humans, and partner systems share regulated routes under declared rules — is what the marker-track primitive supplies.
Vendor and Product Reality
Ocado Group is a UK-listed technology and retail company whose Ocado Smart Platform (OSP) is licensed to grocery retailers globally as an end-to-end e-commerce, fulfillment, and last-mile solution. The physical heart of OSP is the Customer Fulfillment Centre (CFC), a purpose-built warehouse organized around a three-dimensional aluminium grid known as The Hive. On top of the grid, fleets of Ocado Storage and Goods Vehicles (SGVs) — small, wheeled, communicating bots — move continuously to retrieve totes containing individual products, present them to picking stations, and replace them. A typical large CFC operates several thousand bots simultaneously on a tightly synchronized grid running at sub-second control cadence.
The platform is licensed to Kroger in the United States, Aeon in Japan, Sobeys in Canada, Casino in France, Coles in Australia, ICA in Sweden, and others. The Re:Imagined program is Ocado's next-generation hardware and software refresh: lighter bots, denser grids, faster picking robots, and a modular CFC footprint that supports smaller "zoom" sites alongside flagship facilities. Adjacent product lines include Ocado Intelligent Automation for non-grocery industrial customers and last-mile routing software for the delivery vehicle fleet.
Architectural Gap
OSP is, today, a closed and exquisitely tuned single-operator system. Every bot on a given grid is owned, scheduled, and controlled by Ocado's central control plane. That works beautifully inside a CFC. It does not extend cleanly to the world outside the grid, and increasingly Ocado's customers want exactly that extension. A Kroger CFC sits next to a Kroger conventional distribution center with its own forklifts, drivers, and inventory systems. An Aeon site sits inside a regulated Japanese logistics corridor that contains other operators' vehicles. A Re:Imagined zoom site sits inside an existing supermarket where store associates, customers, and third-party delivery riders share floor space with bots.
Each of these adjacencies is a lane-authority problem. Who is allowed in which corridor at which time? Under whose credential? With what fallback when a human enters a robot lane or a robot strays into a regulated human-only zone? OSP solves this inside the grid by simply owning every actor. Outside the grid, the pattern breaks. The current workaround is hard physical separation — fences, doors, dedicated time windows — which throws away most of the density benefit of automation. The architectural gap is the absence of a credentialed routing substrate that can carry mixed-fleet, mixed-operator, mixed-human traffic under declared rules.
What the AQ Marker-Track Primitive Provides
Marker-track is the Adaptive Query primitive for regulated, credentialed routing across shared physical or logical lanes. Each lane — a corridor on a warehouse floor, a slot on a conveyor, a delivery curb cut, a transfer aisle between two operators' sites — is a declared resource with an authority and a policy. Each mover — an Ocado SGV, a Kroger forklift, an Aeon store associate, a third-party delivery rider — carries credentials that bind it to the lanes it is permitted to traverse and the conditions under which it may do so. The substrate composes movers and lanes into routes whose every transition is signed, replayable, and auditable.
The primitive treats authority as the first-class object, not the mover. That is the structural difference from current warehouse-control systems. A bot does not have intrinsic right-of-way; it has credentials that the lane recognizes. When credentials change — a corridor enters a maintenance window, a human enters a normally robot-only lane, a partner operator's shift starts — routes are recomposed under the new authority state without re-architecting the control plane.
Composition Pathway
Ocado integrates as a high-density credentialed operator inside the substrate. The CFC grid itself becomes a declared lane network with Ocado as the issuing authority for all internal movers; this preserves OSP's sub-second control discipline because the substrate does not interfere with the grid's internal scheduling, it only formalizes the grid as one credentialed region in a larger map. At the boundary — receiving docks, transfer aisles, store-floor zoom sites, last-mile staging — the grid's authority gives way to a partner authority (Kroger's, Aeon's, a municipal curb regulator's) and the substrate handles the credential handoff.
Mixed-human zones become tractable. A store associate's badge is a credential; a customer's presence is a transient anonymous credential with conservative routing rules; a third-party delivery rider's credential is issued by a partner authority Ocado recognizes. Re:Imagined zoom sites benefit most directly because they explicitly target shared store environments where hard physical separation is not acceptable.
Commercial Position
Ocado's commercial position improves along the dimension its customers actually push on: extending automation past the CFC boundary without giving up safety or regulatory standing. Today, every conversation with Kroger, Aeon, or a new licensee about expanding into store-floor automation, dark-store conversions, or shared logistics corridors hits the same wall — OSP can run the bots, but it cannot speak to the partner systems and human workflows that share the space. Marker-track participation removes that wall. Ocado retains its grid IP, its bot fleet, and its software-licensing revenue model, and gains a position in mixed-environment automation that is otherwise inaccessible.
The position also matters defensively. Competing automation vendors and emerging warehouse-robotics entrants are pitching openness and interoperability as a wedge against OSP's closed-system reputation. Marker-track participation lets Ocado answer that pitch architecturally rather than rhetorically.
Licensing Implication
Marker-track is licensed at the lane-authority and credential-issuer level. Ocado licenses an authority role for each CFC and each zoom site, plus credential issuance for its bot fleet. Partner operators license their own authorities for the zones they own. Cross-authority routing — the actual point of the primitive — is licensed at the substrate, not negotiated bilaterally between operators. For Ocado this is favorable: it preserves grid sovereignty inside the CFC, formalizes a clean handoff at the boundary, and turns what is currently a custom-integration cost center into a declared, replayable, regulator-presentable operating model. Safety-case documentation, which retailers and labor regulators increasingly demand for shared human-robot environments, is produced as a byproduct of substrate operation rather than as a separate compliance artifact.