How Meshmerize Keeps 32 Autonomous Shuttles Connected Across 15 Levels — With Zero Handover Time ?
There are no windows. No lights. No people walking the floor. By design, the warehouse runs in complete darkness — because it does not need humans inside to operate.
This fully automated facility stands 50 metres tall, holds over 80,000 pallets each weighing up to one ton, and moves 12,000 pallets every 24 hours across 15 storage levels. Every pallet is handled by one of 32 autonomous shuttles, controlled remotely, moving at speeds of up to 4 m/s through dense metal racking from floor to ceiling.
It is one of the largest autonomous warehouses in the world. And keeping it running depends entirely on the network.
Meet the Atlas Shuttle?
Movu, a member of the Stow Group, built the Atlas shuttle specifically for extreme warehouse conditions. It operates in temperatures ranging from -25°C to +45°C, carries loads of up to 1,500 kg, and reaches a top speed of 1.4 m/s during normal operations.
Furthermore, its deep lane access design reduces the need for wide driving aisles — maximising storage density in a facility where every centimetre counts. Atlas is not just fast and robust. It is built to run continuously, at scale, without interruption.
That last requirement is where the network becomes critical.
Deployment Detail |
Figure |
|---|---|
| Warehouse height | 50 m |
| Storage levels | 15 |
| Pallet capacity | 80,000+ |
| Pallets moved per day | 12,000 |
| Active shuttles | 32 |
| Static access points | 66 |
| Shuttle top speed | 4 m/s |
| Safety network avg. ping | ~10ms |
| Handover time | 0ms |
The Scale of the Challenge ?
The warehouse runs 66 static wireless access points, connected by ethernet cable and positioned an average of 20 metres apart across three distinct floors. Each access point uses a Doodle Labs RM-2450 device fitted with omnidirectional antennas. The 32 shuttles carry the same hardware, running Meshmerize software and fitted with specialist antennas suited to the environment.
Every 100 milliseconds, the central traffic controller sends an MQTT command to each shuttle. Each shuttle responds immediately, confirming its position and status. If a shuttle fails to respond within 2 seconds of receiving a command, the system registers it as offline and triggers a disconnect state.
Two seconds sounds generous. In this environment, however, it is razor thin.
Why Standard Wi-Fi Cannot Handle This ?
Each shuttle crosses a static access point roughly every 10 seconds during normal operations. Metal racking surrounds them on all sides. Liquids sometimes travel on top of them. The physical environment creates constant radio interference and shifting dead zones.
With conventional Wi-Fi roaming, handover between access points takes between 50ms and 500ms under normal conditions — and up to 2 seconds in difficult environments. Consequently, each shuttle would experience multiple disconnections every single minute. In a fully automated facility with no staff on the floor to intervene, a single disconnection cascade could bring the entire warehouse to a halt.
How Meshmerize Solves It ?
Rather than routing each shuttle through a single access point at a time, Meshmerize connects every shuttle to multiple nodes simultaneously. With approximately 100 nodes and thousands of active links between them, the multipath routing engine continuously calculates the most reliable path for every packet — and reroutes instantly when conditions change.
As a result, handover time drops to zero. Shuttles move continuously across all 15 levels without a single roaming gap. The average ping time on the safety network runs at around 10ms with very low jitter.
The raw MQTT data bandwidth sits at around 50 kbps per shuttle. However, when network headers and overhead are added across 32 active shuttles, total network throughput reaches 200 to 250 kbps — all of it managed without congestion or priority conflict.
Two Networks, One Infrastructure
Not all traffic in the warehouse carries the same urgency. Therefore, Meshmerize separates it into two distinct paths using the dual radio setup on each device.
The 5 GHz network carries all latency-critical traffic — safety signals, shuttle control commands, and keep-alive messages. The 2.4 GHz network handles everything else: logging data, remote access, and maintenance traffic that needs bandwidth but tolerates higher latency.
Additionally, maintenance engineers can connect directly to an individual shuttle’s onboard hotspot using a smartphone — allowing manual control during interventions without disrupting the rest of the network.
When a new node joins the network, staff simply assign it a unique name. The traffic controller picks it up automatically. No manual configuration is required.
The Bottom Line
This deployment has been in full-scale production since spring 2023. It stands as one of the most densely populated dynamic mesh networks in the world — and it runs without interruption, in complete darkness, every day.
For a full technical breakdown of the deployment, you can read the complete case study here.
- Meshmerize is based in Dresden, Germany. To find out what this looks like in your facility, reach out at hello@meshmerize.net.
