Mapping the Future
Sept. 16, 2020—From navigating the family hiking trip to choosing which neighborhood has the best school districts—it wouldn’t be possible without a map. The first maps have been traced back as early as 14,500 BCE. They were found drawn on the inside of caves and depicted water sources, hunting grounds, and detailed sketches of the stars in the night sky. The cartographers of today do not rely on visibility as they once did. Instead, they rely on data—and lots of it.
Partners for Automated Vehicle Education held a webinar earlier this month to discuss the role maps play in informing autonomous vehicles. One of the panelists was Founder and CEO of CARMERA, Ro Gupta. First and foremost, maps happen in layers, Gupta says.
“How do you layer data on top of a map that makes it more useful?” poses Avery Ash, head of autonomous mobility at INRIX, a private company that provides location-based data and analytics.
CARMERA on the other hand, is a private company that builds high-definition, fully regenerative maps for autonomous vehicles, and aggregates its data from crowd-sourced cameras and sensors. When it comes to compiling that data, CARMERA knows just where to start.
Breaking it Down
Gupta says the first step is localization. Localization is essentially answering the AVs question of “Where exactly am I?” he says. If the AV is unable to locate itself, it won’t be able to locate much else, let alone traverse through unexpected environments.
Next comes perception. Perception refers to how the AV is interpreting its surrounding environment. An example of this would be how the AV operates when it is placed in front of a sandy beach versus, say, a billboard featuring a sandy beach.
Gupta says this answers the question, “Am I seeing what I think I’m seeing?” for the AV.
The final piece of the puzzle is planning. Gupta says HD maps help the AVs plan for the road ahead by anticipating risks and providing alternate routes should they arise.
Mapping the City
Brittney Kohler is the director of transportation and infrastructure at National League of Cities, an advocacy organization that represents the country’s cities, towns and villages.
Kohler says at NLC, they have been talking about the mapping world as a part of city function, but also city future.
“Does the infrastructure need to change to match technology or should technology change to match the pace of infrastructure?,” she asks rhetorically . “They need to meet each other in the middle.”
INRIX has been discussing the difference between physical versus digital infrastructure, says Ash. Physical infrastructure refers to stop lights, speed limit signs, basically anything a human driver would look to for direction.
Digital infrastructure is not visible to the human eye, only to the vehicle’s computer system.
“When a person sees an eight-sided metal sign, they know it’s a stop sign,” says Ash. “But the car doesn’t know what that means.”
The coding and software put into the car dictate how it perceives and reacts to the outside world. The software acts as a brain and the cameras and sensors throughout the vehicle are its eyes. When the vehicle encounters a stop sign, the software (or brain) informs the rest of the system that the red and white octagonal structure with the letters S-T-O-P literally means “stop.”
Looking to the Future
The concept of a map may seem archaic, but for an automated vehicle everything is uncharted territory. The aggregation of data is the sole catalyst in the evolution of mapmaking. The more data that is collected, the more diversified and detailed maps come out of it. When it comes to the development of maps for autonomous vehicles, Ash has a few ideas of what could be around the corner.
“We could embed information into road paint or put QR codes on stop signs,” he says. “It’s all about what the vehicle needs.”
Whether digital infrastructure merges with physical infrastructure, or overtakes it entirely—Gupta says it is ultimately up to privately-held companies.
“We can move faster and we have the tools to be able to do it,” he says of the private sector.