Technology / #ForbesTechnology

March 21, 2019,   4:43 PM

Dirty Work For Robots: How Two Young Immigrant Entrepreneurs Created Oven-Sized Roombas To Chase A $5B Opportunity

Amy Feldman



Image source: Wikimedia Commons

On a recent late-winter afternoon, 31-year-old Faizan Sheikh, cofounder and chief executive officer of Avidbots, shows off what the company’s 1,050-pound cleaning robot, Neo, can do. Inside an enclosure made of thousands of giant white, blue and yellow Lego bricks at the company’s factory in Kitchener, Ontario, the robot goes through a quality assurance test, turning corners and scrubbing the floor as it goes. Lights along the bottom of the machine glow blue as it moves autonomously, then turn fuchsia as it stops to remap its route, aided by artificial intelligence. Once the machine is fully vetted in the Lego test zone, it will be loaded into a wooden crate and lined up with a handful of others that are waiting to be shipped to customers. Walking the floor, Sheikh, wearing a gray puffy jacket over a blue Avidbots T-shirt, can barely contain his excitement. “Avidbots,” he says, “is like a fantasy space for engineers.”

Autonomous vehicles and robotic food-service startups have captured the public’s imagination, but cleaning robots have enormous, if unsexy, potential. The market for commercial-cleaning equipment is giant and fragmented, accounting for an estimated $5 billion in global sales annually. Sheikh and his cofounder Pablo Molina, also 31, believe that as much as a third of all commercial-cleaning equipment could be autonomous by 2023, giving them an opportunity to build a very large business.

It’s already a pretty valuable one. A new round of funding led by Palo Alto, California-based True Ventures brings the company’s total funding to $36 million, at a valuation that Forbes estimates at roughly $75 million. Since its inception five years ago, Avidbots has sold a few hundred robots at about $50,000 a pop. The machines are toiling away at Changi airport in Singapore, Charles de Gaulle Airport in Paris, Toronto’s Eaton Centre shopping mall and the Rochester Institute of Technology. Revenue last year was some $3 million, but as Avidbots cranks up its sales and production, it could reach $20 million this year.

That’s still a smidge compared with the sales of industry giants like Minneapolis-based Tennant (2018 sales: $1.1 billion), Germany’s Kaercher (more than $2.8 billion in sales) and Denmark’s Nilfisk ($1.2 billion), which churn out cleaning machines that human workers can push or ride. But macroeconomic forces are nudging the industry toward robotics. Building owners and commercial cleaning companies face rising labor costs, including in the U.S. where there is a national movement toward a $15-an-hour minimum wage. They also have trouble finding and holding onto workers. Meanwhile, the cost of sensors, specialized laser systems and cameras required to make robots have come down, while their quality has improved. Assuming those trends continue, business owners will choose to augment labor with capital. Avidbots hopes to benefit.

The economics are already in their favor in many developed countries. Neo costs around $50,000 (the exact price is based on functionality), runs for around four hours on a single charge and should last five years. A typical service plan costs some $500 per month, adding another $6,000 a year. By contrast, a commercial-grade rider-style scrubber that a laborer would operate costs roughly $15,000 up front. That means the additional cost of a cleaning robot is some $35,000 before accounting for maintenance. Labor costs vary based on geography, but on average, U.S. building cleaning workers are paid $27,000 a year, according to the Bureau of Labor Statistics, meaning the investment in a robot should pay off in two years. “When you buy a robot, you get your payback in about a year if you use it 20 hours a week,” Sheikh says. “Depending on the labor rate, it might be more like 18 to 24 months.”

The publicly traded Canadian cleaning firm GDI Integrated Facility Services bought its first robot from Avidbots two years ago and sent it to Eaton Centre, where it scrubbed the floors overnight while the stores were closed. Craig Rudin, the head of GDI’s janitorial supplies business, says the company, which employs 25,000 workers, has since purchased five more Neo robots and set them up at other shopping malls, freeing workers to focus on cleaning bathrooms and garbage areas. That, he says, helps GDI deal with both employee absenteeism, which runs 7% to 10% a day, and high turnover. Each robot deployed for seven hours in a 24-hour period, he figures, can reduce labor needs by 0.7 to 0.8 of a full-time-equivalent employee, a number that will rise along with the bots’ increased battery time and efficiency. His ultimate goal is a fleet of 1,000 floor-cleaning robots that would be spread throughout GDI’s clients’ malls, universities and warehouses that would be operated from a command center. “It’s nice to have shiny objects,” Rudin says. “But for us, we have to make the economics work.”

Sheikh was born in Lahore, Pakistan, and moved to Canada at age 14 as his parents—an electrical engineer and a college lecturer back home—sought a better life for him and his siblings. Molina, the son of a banker and an accountant, grew up in Cuenca, Ecuador, where he played around with technology as a kid and spent countless hours watching The Mobile Cop Jiban, a Japanese television series about a robotic detective who has a variety of autonomous vehicles. He moved to Canada for college, hoping to get better access to emerging technologies.

The two met as freshmen in the robotics program at Canada’s elite University of Waterloo in 2006. Almost immediately, they started talking about their dreams of a robot-enhanced future. “I wanted to build robots and see them in the real world,” Molina says. “That’s how I got together with Faizan.”

After college, they wanted to start a robotics company, but Sheikh needed to get a job because of his obligations, as the eldest son, to help support his family financially. He moved to Ottawa to work as a software engineer at Bridgewater Systems (subsequently acquired by Amdocs). Molina moved there as well to work on a lunar rover project funded by the Canadian space agency, and subsequently enrolled in graduate school at Ottawa’s Carlton University. “One day Pablo came to me, and he said, ‘Faizan, with all the advances that are happening in robotics research, I think now is the time for something to be commercialized, for something to take off, and let’s work on it together,’” Sheikh recalls.

Molina proposed building a snow removal robot, a seemingly obvious idea for two entrepreneurs living in Ottawa. But they quickly realized that the economics of such a seasonal business wouldn’t work. They switched to indoor cleaning, a year-round operation.

In January 2014, Sheikh quit his job and Molina took a leave of absence from grad school, and the two went to Shenzhen, China, as part of the Hax Accelerator, a respected program for hardware entrepreneurs, where they spent six months and received $25,000 in funding. It was an exciting place to be, but it was also tough for two entrepreneurs who spoke no Mandarin to navigate the city’s huge electronics markets and get quality parts. “You spend a giant amount of time debugging and wondering why it’s not working,” Sheikh recalls.

The duo started building prototypes in China, at first taking the cheapest cleaning machines they could find and retrofitting them as robots. Their first attempt, called Sweepy, was a yellow machine with an off-the-shelf BattleKit motor attached on top. The second was a retrofitted gray scrubbing machine called Scrubby. Hacking together existing machines with sensors was easier and cheaper than building a robot from the ground up, but the machines were too big and were difficult to control.

Not long after they returned to the U.S. from China, they met Rohit Sharma, a general partner at True Ventures, who told them flat-out: You’re taking the wrong approach and need to build the machines from scratch. “I don’t think they were thrilled by that response,” says Sharma, who passed on the deal then but invested two years later.

They did eventually change their strategy for the machine that became Neo (an homage to the Keanu Reeves character in the 1999 sci-fi movie The Matrix), and they now consider building from scratch a major advantage. As their designs got better, they increased the battery capacity from a meager 45 minutes of running time to around four hours, added separate 32-gallon tanks for the dirty and clean water and developed artificial intelligence that can map out the robot’s path and adjust to floor layout changes.

At the factory, Avidbots keeps a lineup of its early prototypes. The body of the last one in the series is a big wooden box that barely looks like a machine, developed in 2016 as a cost-saving alternative to making a mold for plastics. “We were low on money, so our engineer went to Home Depot and hand-built this thing,” says Sheikh, who considers frugality a core tenet of the company. “Today I can romanticize that period a lot looking back on it.”

All told, Avidbots made nine prototypes. The sixth was sent into the world for a two-week test at Mohawk College in Ontario, where it cleaned successfully. Then a Singaporean distributor picked up the Neo for full-time use at KK Women’s and Children’s Hospital. As Sheikh and Molina went to trade shows to demo their new robot, they started to pick up bigger customers. Their robots are now deployed in 14 countries, including the airports in Paris and Singapore, which has wrapped its robots in a humanoid casing complete with eyes and tuxedo.

Jenny Lee, a managing partner at GGV Capital and member of the Forbes’ Midas List, who invested in the latest funding round, says she believes Avidbots can sell 100 robots a month. “The demand for the product has grown quite substantially, and the demand is global,” she says. “The key is can they deliver fast enough, and make sure that what is out there can perform.” For any hardware company, after all, cranking up production without running into quality glitches is extremely difficult.

Meanwhile, Sheikh and Molina are developing new functions, such as security, for their existing robots that will make them a better investment for customers, while also testing new machines, such as autonomous vacuum cleaners, that they might someday release. “By putting robots into the real world, we are changing the perception around robots,” Sheikh says. “That lends itself to really cool possibilities.”

Recommended Articles