Can You Automate a Warehouse You Are Already Running?

Taitung, a frozen and ambient food supplier in Sydney, went live on freezer-rated automation over one weekend, in a building that was already working. Mushiny's reference sites page lists it as a "brownfield install". Nobody had to build a new shed first.

Most buyers in Australia assume they would have to. They lease an existing building, they cannot shut it for months, and they have heard that automation means a purpose-built facility. For a large part of the range, that assumption is wrong. Mushiny's answer is that "Most Mushiny systems are brownfield. They drop into your existing building."

This article explains why that works, what your building still has to give you, and which buildings are poor candidates. The main example is shelf-to-person, the clearest case of automation that sits on a floor rather than in a structure.

Do you need a new building to automate?

No. The assumption comes from older automation, where the machine and the building were one design.

A goods-to-person grid is different. The pods and the robots stand on the existing floor. That gives it three properties, and they are the whole argument.

It is free-standing. Nothing is built into the building. The pods stand on the slab and the robots drive across it, so there is little or no structural work.

It goes in by section. You take one area, stand the grid up, run it, then take the next. The whole building does not have to be ready on one date.

It can come out. At the end of a lease you take it with you, and the building goes back to its owner without a system built into it.

Mushiny's shelf-to-person page describes the robots as driving "on open floor, not a fixed track, so they route around each other." The three properties follow from that. The building is not part of the machine, so there is no steel to hang from the roof and nothing to cut into the slab.

How does that compare with a shuttle or conveyor retrofit?

They are good systems, and they suit sites where density or throughput justifies the work. They ask more of the building.

A shuttle ASRS runs on racking built for it. Mushiny's own page for its 4-way shuttle says the grid "is a new structure rather than a conversion of what you have." Tall racking brings further demands. As the storage density and height article sets out, racking to 12 metres needs a roof, a slab and a structure that can carry it, plus structural certification and fire protection.

Conveyor has the same problem in a different shape. It is installed into the building, and when your flow changes you change the steel. A fixed structure built for one machine is also hard to relocate, which matters when your lease is short or your layout is uncertain.

A retrofit of either kind is a building project as well as an automation project.

One caution on Mushiny's own range. The free-standing argument is strongest for the shelf-to-person grid. MIX Pro stores totes in racking up to 12 metres high, so the height article applies to it in full.

What does the building have to give you?

Less than most people expect. There are two things that actually matter.

A reasonably flat floor, without large cracks or joins. The robots drive on the slab and read QR markers fixed to it, so the surface matters more than the structure. It does not need to be a new floor or a specially ground one. It needs to be flat enough and sound enough, and large cracks or wide joins are the defects worth looking at.

Height, which shapes the design rather than ruling you out. A taller building lets us go taller and store more in the same footprint. A lower one pushes the design wider. Either way we can work in essentially any building, so height is an input to the solution rather than a gate you pass or fail.

That is close to the whole list. There is no slab rating to chase, no flatness class to specify and no structural work, because a goods-to-person grid stands on your existing floor instead of being built into your building. Pods stand up to 2.7 metres, so allow for sprinklers, lighting and structure when you measure usable height. Beyond that, the practical questions are where goods arrive, where people will stand at stations outside the fenced area, and where power reaches the stations and the chargers. A MIX 4.0 battery runs on 48 V, charges in 2 hours and lasts 12.

One point on why the floor is the thing that matters. Our grid robots navigate by QR markers on the floor, inside a fenced area that people stay out of. QR is faster, cheaper and more accurate than SLAM, the mapping method for robots that share space with people, and it works precisely because the robots are isolated from operators. A marker tells a robot where it is, not where to go. The robot still chooses its own route and negotiates with the others. Because the markers live on the floor, the floor is what we look at first.

How do you go live without stopping dispatch?

Mushiny's answer is that goods-to-person and sortation deployments are "typically commissioned in phases, and several of our Australian sites have gone live over a single weekend with no interruption to daily dispatch."

Taitung is the named case. The references page records it as a "Brownfield install and go live in one weekend", with freezer-rated automation and Reverse Drop Sequencing.

A weekend go-live describes the switch-over, not the preparation before it, and no page says how long that took. Ask what has to happen before the weekend.

The references page does not label any other site as a retrofit. Winc, Munro Footwear Group, Coca-Cola, Alloys and Monde Nissin run Mushiny robots in Australia, but the pages do not say whether each went into a working building. Treat Taitung as the one documented example.

What do you do with the stock while you install?

This is the hard part, and the website does not answer it.

A grid has to stand where stock already sits. The stock has to go somewhere while the area is cleared, then onto pods. Three things decide how hard that is.

Where stock can wait. You need floor for it. A building with a spare bay, a mezzanine or a second site has an easier job than one that is full.

Which stock moves first. Clearing a section means choosing which lines move onto pods and which stay on the old shelving. The order changes how much of the old system you can keep running.

How much you hold. A large holding of slow movers takes longer to move than a lean one.

This is the part buyers underestimate. Ask for the stock-move plan before you ask for the robot count.

Can you run the old and new systems side by side?

Phased commissioning means you do, for a period. While one section runs on robots, the rest of the building keeps picking the old way. Orders may draw from both areas, and your warehouse management system has to handle that.

The overlap has a cost. You run two methods, train people on one while using the other, and carry split stock. The benefit is that dispatch never stops. How long it lasts depends on how many sections you phase the install into, and nobody can say without your layout.

What makes a building a bad candidate?

Fewer buildings than you would guess, and the real problems are rarely the shed itself.

A poor floor. This is the main one. The robots drive on the slab and read QR markers fixed to it, so large cracks, wide joins or a broken surface are a reason to stop and survey before anything else is decided.

A very low roof. Height does not rule you out, but it does shape what you get. Pods stand up to 2.7 metres and sprinklers, lighting and structure all take height, so a low building pushes the design wider instead of taller and a tall tote store such as MIX Pro stops making sense.

A tight column grid. Columns interrupt a grid. It is usually a design problem rather than a stopper, but if columns chop the area into small pieces there may be no section large enough to be worth standing a grid up in.

Nowhere to stage. If you cannot move stock out of an area, you cannot clear it. A full building with no spare floor and no way to free a section is a hard install.

Nothing to fix. If walking is not your constraint, a grid is the wrong tool in any building. A lease ending in months may also be too short to recover the cost.

What should you gather before you call anyone?

1. Your building drawings, with clear height, column positions and doors marked.
2. The floor's condition, and photographs of any large cracks or wide joins.
3. The first area you could free, and where its stock could wait.
4. Your lease end date and any rules on altering the premises.
5. Where the time goes today, measured with a stopwatch for one shift.

Bring them to the Chatswood Tech Centre, the only place in Australia with Mushiny's full range running live under one roof. It is free to visit.

Frequently asked questions