Carmera, the maker of dynamic, real-time, high-definition maps for use in mobility applications, announced today that it has entered into a commercial partnership with Toyota Research Institute-Advanced Development, the carmaker’s division focused on the development of autonomous vehicle technologies. The monetary value of the partnership was not disclosed.
The two companies will collaborate on a proof-of-concept project to develop “camera-based automation of HD maps” for urban and surface roads—a more challenging environment to map than highways, Carmera and TRI said in a press release.
Ro Gupta, CEO of Carmera, says the project is the first step in realizing Toyota’s new open-source Automated Mapping Platform (AMP). Although the platform is in its infancy, AMP seeks to accelerate the scalability of self-driving cars by generating automated HD maps, which could help AVs navigate, from the combined data gathered from participating companies.
Mapping small areas is a noble pursuit, but Gupta says Toyota’s goal of providing a global safety infrastructure for autonomous driving through AMP and keeping those maps updated is one reason the partnership appealed to his company.
“What’s important, and why we’re excited about the partnership, is that TRI-AD is very focused on production and scalability,” Gupta explains. “All of its work anchors around that central theme. It’s one thing to do demos and research in small, confined areas. To scale, you have to address a number of environmental and economic challenges.”
Carmera’s crowdsourced, real-time maps help autonomous vehicles know where they are at any given moment. To create the maps, the company incorporates continually updating sensor data from a network of commercial fleet vehicles. Carmera’s maps also assist with path planning, which allow AVs to predict what kinds of roadways and obstacles to expect in advance.
For the TRI-AD mapping project, the two companies will install cameras in Toyota test vehicles to collect data over several months in downtown Tokyo, Japan. That data will then be processed on Carmera’s platform to automatically generate HD maps.
Static base maps of highways for use by autonomous vehicles are already being developed by a number of companies, although Gupta feels the mobility industry has progressed enough that completing those maps is more of a “soft problem.”
“Building a map once is hard, but it’s even harder to keep it updated, especially if you require accuracy and a fast turnaround time,” he says.
Autonomous vehicles are trying to answer three questions: where am I, what am I seeing, and where do I go next—known in the industry as localization, perception, and planning. Gupta says TRI-AD is relying heavily on maps to answer those questions. Cameras are cheaper than other sensing technologies, such as LiDAR, and they can work in concert with other sensor data to create a complete picture of the car’s surrounding environment, he adds.
Carmera, which was founded in 2015, has already gathered and processed videos documenting millions of miles of driving from partner sources to build its mapping platform, Gupta notes. The company is based in New York City and Seattle in part because it wanted to start with mapping the complex urban environments those cities offer.
Carmera had a busy year in 2018. In addition to initiating several mobility partnerships with auto and tech companies, it raised a $20 million Series B funding round over the summer that was led by GV (formerly Google Ventures). To get buy-in from Google, Toyota, and other industry leaders has been “absolutely validating,” Gupta says.
In addition to New York and Seattle, the 45-employee company has satellite offices in Detroit, San Francisco, Japan, South Korea, and the United Kingdom.
Toyota Research Institute, which works in collaboration with nearby universities, has locations in Ann Arbor, MI; Cambridge, MA; and Palo Alto, CA.
