An Uber self-driving car drives down 5th Street in San Francisco, California. Cars in Uber's self-driving cars are back on the roads after the program was temporarily halted following a crash in Tempe, Arizona in March. (Pic: Justin Sullivan, AFP)
Washington - A
mile away from where Google builds the maps used by people around the
world, a 25-person startup is trying to do something similar for robots.
DeepMap, which was founded by mapping veterans of Alphabet, is building systems enabling self-driving cars to steer through
complex cityscapes. DeepMap plans to license its map-building software
to automakers and technology companies looking to teach cars how to
The company recently raised a $25m funding round led by Accel,
adding to the $7m it raised last year. In the chase to secure
the investment, the firm penned an 80-page white paper on DeepMap's
Investors are eager to grab a piece of self-driving cars after big
startup deals, such as General Motors' acquisition of Cruise
Automation and Uber Technologies' purchase of Otto. But new
entrants face sizable competition, whether in autonomous technology or
mapping. In addition to Silicon Valley juggernauts Alphabet, Apple
and Uber, Amsterdam's TomTom is building its own 3-D maps, as is
HERE, a project jointly owned by several automotive giants and Intel.
Andreessen Horowitz and GSR Ventures also invested in DeepMap. Martin
Casado, the lead Andreessen partner on DeepMap, said startups can
compete with much larger companies because their businesses have fewer
conflicts. This enables them to work with just about anybody and focus
on the software, which typically has higher profit margins. "DeepMap
isn't bound to a particular application like ride-sharing, and they're
not bound to a particular company," Casado said.
DeepMap fuses images from digital cameras with data collected using a
laser device known as lidar to build detailed 3-D maps. It can identify
street signs, billboards and the height of a curb. The 3-D maps
complement the sensors by giving the cars a detailed awareness of the
environment outside the car's view. James Wu, who started the company
with his former Google Maps colleague Mark Wheeler, said he hopes to
have three customers by the end of the year, while generating as much as
$10m in revenue.
"Having the map lets us define for the robot the rules of the world,"
said Wu, who has also worked on mapping projects at Apple and Baidu.
"It's very easy to make a prototype car that can make a few decisions
around a few blocks, but it's harder when you get out into the world."
A diaspora of autonomous driving entrepreneurs from Google has spread
throughout Silicon Valley. Google X founder Sebastian Thrun just spun
an autonomous car startup out of his online learning company Udacity.
Top engineers Chris Urmson and Bryan Salesky have set up companies
called Aurora Innovation and Argo AI, respectively.
Anthony Levandowski, is the focus of an embittered legal battle between
his former employer Alphabet and Uber over allegations that he stole
intellectual property from the search giant when he left to form his
startup, Otto.Read Fin24's top stories trending on Twitter: