A practical readiness guide for ramps, traffic routes, safety controls, charging, maintenance and data-led scaling.

Professional illustration of autonomous electric warehouse vehicles moving goods between floors on a ramp, with facility-readiness callouts for traffic routes, charging, staging and safety.

Autonomous vehicles are moving from controlled demonstrations towards more practical warehouse applications. For Singapore facility managers and logistics operators, the important question is not only whether an autonomous vehicle can move a pallet. It is whether the building, operating procedures and people around it are ready for safe and reliable scaling.

CEVA Logistics’ announcement on 10 September 2026 provides a timely local case study. CEVA is testing two Zelostech Z10 autonomous electric vehicles at its Singapore Blue Hub. The vehicles are being used for intra-hub transfers involving pallets, totes and other inventory across different floors, using an existing multi-level ramp. CEVA states that the pilot is intended to improve operational agility, staging-area utilisation and scheduling flexibility, with an expected reduction of at least 4,600 kilograms of CO2 annually, subject to trial results.

This remains a pilot, not evidence that every multi-level warehouse is ready for autonomous vehicle deployment. However, it shows why engineering and facility-management teams should assess the operating environment before expanding from a small trial to a larger fleet.

1. Assess the building route, not just the vehicle

A multi-level warehouse introduces physical constraints that do not exist in a single-floor facility. Before deployment, operators should map the complete vehicle route from pickup to drop-off, including ramps, turning points, loading zones, crossings, lift lobbies and temporary obstructions.

Key questions include:

  • Is the ramp wide enough for the vehicle, the load and safe separation from people or other equipment?
  • Are gradients, floor transitions and surface conditions suitable for repeated movement?
  • Can the vehicle maintain reliable navigation at corners, blind spots and changing light levels?
  • Are there sufficient clearance heights and protected areas for pallets, totes and other loads?
  • Will fire doors, roller shutters, access controls or other building systems affect the route?

A route that works during a quiet pilot may behave differently during peak operations. Facility teams should therefore test the route under realistic conditions, including pedestrian activity, forklifts, delivery movements and temporary staging.

2. Design traffic rules for mixed human and robotic movement

Autonomous vehicles rarely operate in isolation. Intra-hub routes may also be used by workers, forklifts, pallet jacks, contractors and visitors. The warehouse needs a clear traffic-management plan that explains who has priority, where people may cross and how vehicles behave when a route is blocked.

Useful controls may include marked pedestrian crossings, restricted access zones, speed controls, warning indicators, one-way sections and designated waiting areas. These should be supported by induction and refresher training rather than relying only on signs or technology.

Operators should also define how the system handles unexpected human behaviour. For example, what happens when a worker leaves a trolley in the route, a pallet is placed outside its assigned location or a person enters a low-visibility area? These scenarios should be tested during commissioning and reviewed after incidents or near misses.

3. Treat staging areas as part of the automation system

CEVA’s pilot is intended to support staging-area utilisation and scheduling flexibility. That makes staging design a central readiness issue, not an afterthought.

Each pickup and drop-off point should have enough room for safe loading, unloading, inspection and exception handling. Operators should consider whether goods can be presented consistently to the vehicle and whether an employee can intervene without standing in the vehicle’s travel path.

Floor markings, location identifiers and operating rules should be consistent across levels. If inventory is placed in different positions from one shift to another, the vehicle may be technically capable but operationally unreliable. A good pilot should test the process as well as the vehicle: arrival, load confirmation, transfer, handover, exception handling and return to service.

4. Plan charging and electrical support early

Electric autonomous vehicles require a charging strategy that fits the warehouse’s operating cycle. Charging points should be located where they do not obstruct emergency routes, loading activities or pedestrian circulation. The area should also support inspection, housekeeping and controlled access.

Before installation, building owners and operators should confirm available electrical capacity, equipment location, ventilation needs, cable protection and access for maintenance. The operating plan should define when vehicles charge, how low battery conditions are handled and what happens if a charging point is unavailable.

Charging demand should be measured against actual duty cycles rather than assumed utilisation. A small pilot may operate comfortably with manual scheduling, while a larger deployment may require additional charging capacity, monitoring and contingency planning.

5. Define emergency and recovery procedures

Automation does not remove the need for human decision-making. A vehicle may stop because of an obstruction, sensor issue, communication loss, low battery, damaged load or building-system interruption. Staff need clear instructions for isolating the area, contacting the responsible person and recovering the vehicle safely.

Procedures should cover emergency stops, manual recovery, blocked ramps, collisions, dropped loads, power interruptions and evacuation conditions. They should identify who can authorise a restart and how the incident is recorded.

Singapore’s Workplace Safety and Health technology resources highlight the broader direction towards technology-enabled safety controls, including vehicular-safety technologies and digital work processes. For warehouse operators, the practical priority is to integrate autonomous vehicle procedures into the site’s existing risk assessments, permit controls, incident reporting and emergency arrangements. Technology should strengthen these systems, not operate separately from them.

6. Assign maintenance responsibilities before scaling

A pilot can expose a gap between the vehicle supplier, warehouse operator, building owner and facilities team. Before expanding, responsibilities should be documented for preventive maintenance, software updates, batteries, sensors, tyres, route markers, charging equipment and building interfaces.

Operators should also agree on service response times, spare equipment, escalation contacts and temporary manual operating procedures. A vehicle that is technically available but waiting for an unclear approval or repair process can reduce productivity and create unsafe workarounds.

Routine inspections should include both the vehicle and the environment. Ramps, floor surfaces, lighting, barriers, doors and staging locations can change over time. Facilities teams should have a process for notifying the automation team before layout changes, renovation works or temporary closures affect an autonomous route.

7. Collect data that supports a scale decision

The purpose of a pilot is to produce evidence for the next decision. Useful measures may include completed transfers, travel time, waiting time, route interruptions, intervention frequency, battery performance, charging availability, near misses, damaged loads and manual recovery events.

Data should be reviewed alongside operational outcomes. For example, a reduction in vehicle travel time may not improve productivity if staging congestion increases. Similarly, an emissions estimate should be checked against measured operating patterns, energy use and the scope of the trial. CEVA’s stated expectation of at least 4,600 kilograms of annual CO2 reduction is subject to the pilot results, so operators should treat emissions outcomes as a measurement exercise rather than a guaranteed result.

Before scaling, management should agree on thresholds for safety, reliability, utilisation, service continuity and financial performance. The decision should also consider workforce impact, including revised roles, supervision, training and opportunities for employees to move into higher-value coordination or technical activities.

8. Use a staged readiness process

A practical Singapore deployment can begin with a defined route and limited operating window, then expand only after the site demonstrates stable performance. Suggested stages include route assessment, controlled commissioning, mixed-traffic testing, peak-period testing, emergency drills, data review and formal scale approval.

Singapore’s built-environment initiatives are also encouraging the movement of robotics solutions from pilots towards wider adoption. That does not eliminate the need for site-specific engineering or workplace-safety assessment. It reinforces the value of documenting what works, what fails and what must change before a solution is repeated elsewhere.

What warehouse operators should do next

For a multi-level warehouse considering autonomous intra-hub movement, the next step should be a joint review involving operations, facilities, workplace safety, IT or automation specialists and relevant suppliers. Start with one representative route and document the physical, operational and human requirements before selecting a fleet size.

ISS can support Singapore businesses in reviewing facility conditions, engineering interfaces, operational controls and AI automation requirements for warehouse environments. Contact ISS to discuss your engineering, facility management or AI automation needs.

Sources