IN Brief:
- Two Zelostech Z10 autonomous electric vehicles are operating inside CEVA's eight-storey Blue Hub in Singapore.
- The vehicles move pallets, totes, and other inventory between floors using the building's existing multi-level ramp.
- CEVA expects at least 4,600kg of annual CO₂ savings and will use the trial to assess wider deployment.
CEVA Logistics has deployed two autonomous electric vehicles at its Blue Hub in Singapore, extending automation into the movement of pallets, totes, and other inventory between floors. The two Zelostech Z10 vehicles use the eight-storey facility’s existing multi-level ramp, replacing diesel truck movements on selected internal transfers without requiring a dedicated lift or fixed guideway.
The pilot is focused on intra-hub transport, where automation has to cope with changing routes, staging priorities, and mixed traffic rather than a single fixed process. CEVA says the vehicles are intended to improve operational agility, make better use of staging areas, and allow transfers to be scheduled more flexibly as demand shifts through the day. Based on current trial results, the company expects the deployment to reduce carbon dioxide emissions by at least 4,600kg a year.
Blue Hub already provides a heavily automated test environment. The 50,000-square-metre site uses goods-to-person systems, automated guided vehicle picking, and robotics while serving luxury, beauty, retail, and healthcare customers. Adding autonomous cross-floor transport therefore extends automation between established processes, where delays can otherwise build as inventory waits for a manual transfer between storage, picking, and staging areas.
Multi-level warehouses create a particular engineering problem because vertical or ramp movements break the flat-floor assumptions behind many autonomous vehicle deployments. Using an existing ramp could allow an operator to automate transfers without installing conveyors or lifts solely for internal transport, but the vehicle still has to navigate gradients, junctions, people, and other equipment reliably. The commercial case will depend on how much integration and supervision are needed once the pilot moves beyond controlled trial conditions.
Labour use is another part of the calculation. Repetitive transfer runs consume driver time without changing the product itself, yet they remain necessary if pallets or totes have to move between operational zones. An autonomous vehicle can absorb some of that routine movement and leave staff available for exception handling, loading, quality checks, or other tasks, provided warehouse and traffic-control systems can dispatch the vehicle accurately and prevent conflicting movements.
The Singapore project grew out of the 2025 CMA CGM Startup Awards, co-organised by innovation platform ZEBOX. Zelostech received funding to develop a proof of concept with CMA CGM, CEVA’s parent group, before the technology moved into the Blue Hub pilot. The progression from a funded proof of concept to a working warehouse trial gives CEVA a chance to measure the system against live operating demands rather than a demonstration route.
Scaling the technology across CEVA’s wider network would require more than proving that two vehicles can use one building. Facilities differ in layout, traffic density, floor loading, gradients, IT architecture, and the balance between automated and manual work. A system that can be configured around existing infrastructure with limited site-specific engineering is easier to reproduce than one whose business case depends on extensive civil works or bespoke integration.
Electrification and autonomy also alter different parts of the cost base. Replacing diesel propulsion can cut local emissions and fuel use, while autonomous operation changes labour allocation, scheduling, utilisation, maintenance, and software requirements. Combining both means CEVA will need to judge the vehicles on total operating performance rather than the emissions figure alone, particularly if future deployments are expected to run at higher utilisation than the Singapore trial. CEVA also needs to determine how charging, battery availability, and maintenance fit around the existing shift pattern, because an electric vehicle that is unavailable at a peak transfer period can simply move the bottleneck elsewhere in the building.
The 4.6-tonne annual carbon reduction is modest in isolation for a global logistics group, but repeatability determines whether a small site-level saving becomes material. A vehicle architecture that can be used across many multi-level facilities could produce cumulative benefits in fuel use and routine transfer labour; a system that demands heavy adaptation at each site would be harder to justify at scale.
CEVA says the Singapore results will inform potential deployment of autonomous mobility solutions at other facilities. The next evidence will come from normal operating conditions — peak periods, staging conflicts, mixed traffic, maintenance downtime, and exception handling — rather than from the vehicles’ ability to complete a prescribed route. If the system continues to use existing building infrastructure without introducing new bottlenecks, Blue Hub will have established a practical route for extending warehouse automation beyond the flat floor.
