Four years and more than 100 companies later, two Pittsburgh startups are still in the running to score part of the $5 million purse for the IBM Watson AI XPrize, an international competition looking for companies using artificial intelligence to tackle some of the world’s most pressing problems.
Now at just 10 teams, Pittsburgh is the most represented city in the race with two groups advancing to the semifinal round: North Side-based Marinus Analytics, which uses software and facial recognition tools to help law enforcement officials identify sex trafficking victims, and CleanRobotics, a startup in Point Breeze that uses artificial intelligence to sort trash to help promote proper recycling.
XPrize plans to announce the three finalists in mid-February. Those three teams will pitch their product on the TED stage in Vancouver in April to compete for the $3 million first-place prize, the $1 million second-place prize or the $500,000 third-place prize.
“They really have framed this competition as the beginning rather than the end of the process,” said Kenny Chen, an ambassador for AI XPrize and the executive director of Pittsburgh-based Partnership to Advance Responsible Technology, a nonprofit think tank.
The XPrize Foundation, a nonprofit based in Culver City, Calif., has operated 17 competitions since it started in 1994, according to its website. It has awarded more than $140 million in prize purses for competitions focused on topics including health, energy, environment, transportation and safety, and robotics.
The foundation credits its first competition, the Ansari XPrize, with introducing private companies to the space industry. That had been exclusively controlled by the government until that point.
In the last few years, XPrize has begun to shift its focus toward artificial intelligence and using large amounts of data to develop solutions, Mr. Chen said. In other words, it will look for companies using artificial intelligence to develop solutions for specific problems.
For example, the foundation recently announced a Rainforest XPrize designed to encourage companies to use data integration to learn more about the forest and to inspire investment and exploration.
“Artificial intelligence is serving much more of a comprehensive, horizontal level of applicability,” Mr. Chen said.
Marinus Analytics, one of the Pittsburgh semifinalists in the ongoing AI XPrize, is getting to ready to launch a service that will allow its customers to more efficiently and effectively use the large amounts of data already available.
The company already works with law enforcement officials to help identify and recover sex trafficking victims, but now it wants to help them tackle large organized crime groups.
“With the click of a button, they can see a pattern or be alerted to that pattern, whereas with the previous methods they would have to piece together these pieces almost manually,” said Emily Kennedy, president and co-founder of Marinus Analytics. “The goal is to reduce that time and increase the amount of cases they can do and enable small agencies to be able to approach these cases.”
Though it took only about a year to implement, the software development is something Ms. Kennedy has been talking about through several rounds of the XPrize competition.
Each company was required to submit progress reports and plans at checkpoints throughout the competition, which officially will span from June 2016 to April 2020.
“You are often just looking forward so it’s nice to have kind of forced time when you have to look back,” Ms. Kennedy said.
Marinus Analytics has previously received funding from the National Science Foundation, the Defense Advanced Research Projects Agency (or DARPA), and the Bank of New York Mellon Corp. The startup has $150,000 in funding, according to Crunchbase, a portal where companies can self-report rounds of funding. If it were to win, the XPrize funding would go toward new hires, Ms. Kennedy said.
Potentially poised to race neck and neck with Marinus Analytics, CleanRobotics has the same growth mindset. But instead of focusing on people, it’s targeting robots.
Charles Yhap, co-founder and CEO of the startup, said it would like to deploy more TrashBots — robots that scan and categorize an item of waste and then place it in its proper stream, which could be anything from landfill to compost to recycling.
He could not say how many units CleanRobotics has already deployed, but said they are in Pittsburgh, Colorado, North Carolina, San Francisco and Australia. This year, it plans to expand to China and Taiwan.
“The crux of this XPrize is AI for good and to my mind we can only be helpful if we make a lot of TrashBots and get them installed,” Mr. Yhap said. “That’s our mission, that’s how we make a difference.”
The units cost anywhere from $500 to $5,000, comparable to the price of a traditional, commercial waste bin, which usually checks in around a few thousand dollars, Mr. Yhap said.
CleanRobotics has $500,000 in funding, according to Crunchbase.
When the AI XPrize was first ramping up, Mr. Chen — who was then working at startup incubator Ascender — said it didn’t seem like many Pittsburgh startups were aware of the competition. Ascender hosted a hackathon at the beginning of 2017 to bring it to their attention and help raise money for the initial entrance fees.
Now, on top of the AI XPrize, Pittsburgh is carving out space for itself in another XPrize competition — the ANA Avatar XPrize, which challenges companies to create an avatar system that will “deploy the senses, actions and presence of a human to a remote location in real time,” according to the competition’s website.
Out of 77 teams, Pittsburgh has two teams in the running: AvaDynamics and Steel City Avatars. That competition, which won’t wrap up until 2022, has a $10 million purse on the line.
Lauren Rosenblatt: lrosenblatt@post-gazette.com, 412-263-1565.
Correction, posted Feb. 5, 2020: The story has been corrected to reflect the total number of teams participating in the IBM Watson AI XPrize competition since the first round.
First Published: February 5, 2020, 11:51 a.m.
Updated: February 5, 2020, 4:52 p.m.
