German 3D Mapping Company Atlatec Is So Confident, Competitors Can Access Its Data

A small German tech company sent a briefcase-sized box to a college student with the instructions to mount it on top of his car and drive around San Francisco’s Financial District many times. The goal—to use the data captured by the GPS antenna and dual cameras in the box to generate what it believes are superior three-dimensional maps that can be used in autonomous vehicles.

The kicker here, is two-fold. The company, Karlsruhe, Germany-based Atlatec Gmbh eschews the use of Lidar and is making its demonstration maps and data available publicly, despite the intensely competitive nature of the 3D mapping sector.

“We’re putting this out there in a range of formats that people who drive on these roads, potentially for their real world tests, can just download from our website, feed into their tech and really put it to the test,” said Atlatec CEO Henning Lategahn in an interview. “There are not many data sets out there that contain maps of that fidelity and maps of that richness and we thought it’s a good idea to share this with the community.”

Of course the other side of that coin is for Atlatec, which has no offices or facilities in the U.S., to gain exposure to potential customers, especially in automotive and tech hubs Detroit and California’s Silicon Valley.

“In the Bay Area there’s lots of mobility companies that are looking to get their AV tech to take off so there’s not only lots of customers but also lots of potential competitors out there because some companies build their own maps so you have to kind of compete with the do-it-yourself option and therefore it’s interesting for us to put this out there as a proof of concept,” said Atlatec business development manager Tom Dahlström. 

Atlatec’s competition is considerable. It includes Lyft
LYFT
Level 5, a unit of Lyft, which generates 3D maps from data generated from its fleet of rideshare vehicles, also without using Lidar, and Deepmap, which does.

Lategahn says Lidar is not necessary and indeed, insists the fidelity of Atlatec’s 3D maps is much higher, maintaining, “The camera is core. The camera contains tons and tons and tons of data—much more data in an image than a Lidar scan and we found ways to harness this.” 

The process then makes use of artificial intelligence to help extract what Lategahn calls “objects” such as traffic signals in order to determine their presence and exact location. That information plus other captured data is then fed into a computer which generates a a map which Lategahn grades as “very good, but not perfect.” That’s the job of “human annotators” who Lategahn says go back through the data set and correct any errors.

The result is a detailed 3D map at a fraction of the cost of one generated with the use of Lidar, according to Lategahn who says his company can build many boxes a year for a few thousand dollars compared to a bill of materials costing anywhere from $300,000-$1 million for what he terms “traditional 3D mapping companies.” 

An interactive topview map allows the viewer to manipulate it.

Lategahn isn’t worried about putting Atlatec’s technology out there where competitors can grab it saying “we think those competitors which are really keen on getting their hands on our data will find it anyway.” 

Indeed, putting it out there is Atlatec’s way of throwing the gauntlet with the confidence of a feisty new combatant.

“I’d rather go out there and show it to the competition,” said Lategahn, “and then, now it’s your turn.”