No cleaners, no choice? Coffee shops bet on machines to do the dirty work

SINGAPORE: Mr Tommy Ng runs five coffee shops, and each outlet needs at least two cleaners per shift just to keep things running smoothly. Finding people willing to do the job — and stay — has become a constant struggle.

“The turnover is very high because the work is tedious,” said Mr Ng, business development manager at Mr Teh Tarik Eating House. “Every few months, there’s a change.”

He is not alone.

Coffee shop operators across Singapore say cleaning roles are among the hardest to fill and keep, largely because the work is physically demanding and widely seen as unappealing.

“The barrier to entry is low, so workers feel they can leave today and find another job tomorrow,” said Mr Wu Yi Feng, director of Kopiwu, which operates coffee shops in Ang Mo Kio, Clementi and Yishun. “They don’t really cherish the job.”

While the mandatory tray return rule introduced several years ago has helped reduce some of the workload, cleaners are still needed to manage return points, wipe tables, and keep dining areas hygienic — especially during peak hours.

Manpower constraints are compounded by Singapore’s foreign worker quotas, said Mr Andy Hoon, chief executive of the Kim San Leng coffee shop chain. Many operators also prefer to hire cleaners directly instead of outsourcing, as engaging cleaning companies would push up operating costs.

Looking to machines for help

With people increasingly hard to find, some operators are turning to technology.

Over the past decade, they have experimented with everything from tray-return robots to robotic arms designed to wipe tables. Results have been uneven — but the search continues.

The latest idea is perhaps the most unusual yet: a table that cleans itself.

At Nanyang Technological University’s Pioneer Food Court, operator Food Haven has installed a prototype self-cleaning table that refreshes its surface with the press of two buttons.

The four-seater table is covered with a disposable sheet. Once activated, the soiled layer — along with leftover food and spills — is pulled into a compartment at one end, while a fresh sheet rolls out across the surface. The entire process takes about 30 seconds.

Each roll contains 30 sheets, enough to last three to four days. A tablet connected to the system tracks how many sheets remain.

Food Haven director Shane Tay said the idea came directly from staffing pressures. The 300-seat food court currently requires three to four cleaners during peak periods.

“There’s always this struggle to maintain proper hygiene in a large dining area because of manpower issues,” he said. “It’s hard to hire workers — and hard to retain them.”

Mr Tay hopes the tables could eventually cut cleaning manpower needs by half and make the job less physically taxing for those who remain. In theory, once diners return their trays, the next customer can sit down at a clean table almost immediately.

Each prototype currently costs between S$500 and S$800 to produce, although Mr Tay said prices could fall with large-scale manufacturing. Future upgrades could include battery-powered operation and more sustainable materials.

Real-world challenges

In practice, the table is not without flaws.

When the prototypes were tested, larger food scraps often got stuck at the narrow opening leading to the collection compartment. Liquids occasionally seeped inside, leaving behind messes that still required manual cleaning.

The table’s bulky ends also take up space, reducing seating capacity. Some diners mistook it for a hotpot or barbecue table, and many were unaware it could clean itself.

Students said dirty tables are a familiar sight during lunchtime rushes at the open-air food court, where leftovers, drink condensation — and even bird droppings — are common. Some said they routinely bring wet wipes to clean tables themselves.

Undergraduate Chua Jia Xing said the design could help improve hygiene but was worried about the amount of waste generated by disposable sheets and possible tripping hazards from exposed power cables.

For cleaner Mdm Vasantha, however, the table has made daily work easier. She said it is simple to operate and requires less scrubbing than traditional tables, easing the physical strain.

Other operators remain cautious.

“This kind of technology can break down,” said Mr Ng. “If it spoils, someone still has to go there and clean the table.”

Mr Hoon questioned whether used rolls might produce odours if left for too long, while Mr Wu raised environmental concerns and noted that fixed designs may not work for shops that use different table shapes.

Not the first experiment

Food Haven’s table is the latest in a long line of automation trials at food centres.

Tray-return robots appeared as early as 2016, before tray return became mandatory, but they were criticised for blocking walkways and still needing to be cleaned. Once fixed return points were introduced, the robots quickly fell out of use.

Table-cleaning robots were also tested, including mobile units equipped with robotic arms. Weston Robot, which developed one such system, said in 2022 that the machines could detect items on tables — but further research and development was needed.

Why cleaning is hard for robots

Experts say food centres are among the toughest environments for robots to operate in.

“These spaces are a super challenge,” said Associate Professor Harold Soh from the National University of Singapore. “Cleaning requires physical intelligence — dexterity, balance — and social intelligence, like knowing when to move and when to stop.”

Robots still struggle with irregular objects, slippery food waste and unpredictable human behaviour, he added.

Associate Professor Lyu Chen from NTU said most commercial robots rely on pre-programmed routines. While navigation has improved, table cleaning demands complex judgment — from recognising spills to distinguishing trash from personal belongings.

“In public spaces, human-robot interaction is not optional,” he said. “The system must behave safely and smoothly around people.”

Automation can help with routine tasks like tray collection or back-end transport, experts said, but human supervision remains essential.

No easy answers

For now, operators say there is no technology ready to replace cleaners entirely — and manpower is still critical.

The labour crunch is also unlikely to ease soon. Mr Wu pointed to Singapore’s ageing population, while Mr Ng noted that many cleaners are retirees who may no longer be able to cope with physically demanding work.

Dr Ng Boon Yuen from the Singapore University of Social Sciences said technology works best when cleaners and operators are involved in shaping it from the start.

“If adoption is low, it’s worth asking whether the technology really reflects users’ needs,” she said.

Researchers continue to push forward on “embodied AI” — robots that can better understand and interact with the physical world. But as Prof Soh put it, bringing robots out of the lab and into everyday coffee shops remains a work in progress.