Where are all the automated warehouses?

Total mechanisation remains a rarity, with some attempts resulting in costly debacles. As businesses find ever more uses for AI, it’s time to ask why robotic repositories haven’t caught on

Illustration of a robot-manned warehouse

Oops. Streetwear retailer End Clothing recently had to announce a £12m stock write-off after a warehouse upgrade went disastrously awry. A technical failure in its shiny new automated fulfilment system had had “adverse effects on both our operations and our customers’ ordering experience”, the firm admitted in a statement.

That’s not the first expensive foul-up to happen in an automated warehouse. In 2019, for instance, Ocado lost an entire facility in a fire caused by an electrical fault affecting one of its robots. The online delivery company estimated that the destruction of its distribution centre in Andover, Hampshire, cost the business more than £100m. 

We’ve been told for years that warehouses are ripe for total automation – robots will fetch and carry everything; radio-frequency identification tags will log goods’ movements; humans won’t be needed. Yet that still hasn’t happened to any significant extent. What’s stopping the march of technology in this part of the business world?

“Despite the clear need for automation and its repeatedly proven benefits, the opportunity remains largely untapped,” reports Maggie Slowik, global industry director for manufacturing at IFS, a logistics consultancy. 

She says that Ocado is actually one of the few success stories so far. At its state-of-the-art warehouse in Erith, south-east London, 2,000 robots pick 2 million items every day. Yet this level of automation is “difficult to accomplish. We’re still years away from achieving wider and deeper adoption. And, while the robot army has replaced hundreds of jobs, there are still humans involved.”

What are the key technological barriers to implementation?

Part of the problem is that robots are more temperamental than people, in that they’re prone to malfunctioning in anything less than a perfect working environment. 

“The current level of automation requires bespoke buildings with precise finishes, such as completely level floors,” says Sarah Bolton, senior associate specialising in commercial real estate at law firm Taylor Wessing. “Just a few millimetres of deviation can cause problems for robots using laser triangulation to navigate, so it can be difficult to retrofit existing warehouses.”

The energy requirements of full mechanisation are potential problem too, as Bolton explains: “Occupiers looking to automate their warehouses need to be particularly mindful of issues concerning the supply of electricity. Automated warehouses can be very energy-intensive. The National Grid’s apparatus is unable to provide sufficient capacity. Upgrades are taking years and millions of pounds to complete.”

The increased cybersecurity risk posed by automation is another factor dissuading companies from filling their warehouses with robots. It’s no secret that so-called smart devices connected to the internet are particularly vulnerable to ransomware attacks. It’s almost impossible to assess a robot vendor’s capacity for issuing security patches, which could render any automated warehouse a tempting soft target for cybercriminals.

Mechanisation presents a significant legal concern in cases where the facility is leased. If a tenant is planning to automate a warehouse, the landlord may justifiably worry about the building’s energy rating, the increased fire risk and the potential difficulty of reletting it if that lessee leaves. With such risks in mind, it may not consent to the plans. 

The current level of automation requires bespoke buildings with precise finishes, such as completely level floors

“It’s going to be a while before everyone breathes a deep sigh of relief because the robots have arrived,” Bolton says.

Obtaining funding for the appropriate upgrades can be easier said than done too, according to Rhyce Dawson, associate director at supply chain consultancy TMX Transform. 

“Capital availability is a massive barrier for firms seeking fully automated warehousing,” he says. “Companies need substantial finance to invest in end-to-end automation. That’s challenging to find, given the rise in interest rates and other economic challenges.”

Any firm that does incorporate robots into its warehouse will have a capacity-related risk to manage. Robots have a maximum output. If the demand on the facility exceeds this, human input will be required to make up the shortfall. But asking people to work alongside machines is not ideal, particularly from a health and safety perspective. The prospect of metal colliding with flesh is unappealing, to say the least.

Dawson notes that “a fully automated warehouse thrives on predictability and a consistent flow of tasks – a scenario that isn’t always aligned with the dynamic nature of most businesses. Peaks in demand, such as those seen during Black Friday, often necessitate human intervention, which renders full automation less practical.”

How to make an automated warehouse work alongside other tech

Suppose that a company can surpass all these obstacles. It has enough capital; a suitably specified building with an adequate energy supply and an understanding landlord; stable demand for its wares; and world-class cybersecurity provision. Even at this point, it’s likely to face further technical challenges.

One of these, Slowik warns, is that it’s quite possible that even the fanciest robots won’t gel with the rest of your tech stack. 

“Companies often choose best-of-breed applications that don’t integrate seamlessly with their enterprise resource planning systems,” she says. “The risk is that the two solutions run on separate databases and you therefore end up with data duplication, which creates complexity and, eventually, more cost.”

Choosing which warehouse robots to invest in is no simple matter. The latest ones are equipped with machine vision, which gives them a near-human ability to identify products. Bosch recently installed them at its plant in Curitiba, Brazil. They identify 7,000 items a day and have reduced the proportion of false rejects to 5%. 

The key consideration for other warehouse managers is whether the robots they’re thinking of buying are as good as this. If not, will the benefits they promise really outweigh the costs? Automation is brutally expensive. Few companies can afford to purchase the wrong kit. 

Lastly, there’s the issue of risk. A successful investment in warehouse robots could radically enhance your operations and provide a good return over the long term through improved cost-efficiency. Or, as End Clothing found, a glitch-plagued implementation could cost your business dearly and generate embarrassing headlines in the national media. 

So the big question is this: are you feeling lucky?