If you can imagine the potential of a system that uses artificial intelligence to turn LiDAR data into useful models—and gets better at that modeling as it goes—well, you’re not the only one. Civil Maps, who makes just a system, has just received $6.6 million in funding from investors like Modus Ventures, Ford Motor Company, and StartX Stanford.
Civil Maps’ particular brand of artificial intelligence modeling software is built on deep-learning technology. Essentially, instead of writing algorithms to recognize specific types of objects in sensor data, Civil Maps wrote a system that learns how to write optimal algorithms all by itself.
The company originally presented its deep-learning system at SPAR 2015, and it’s safe to say that in the intervening time, the system has taught itself to recognize objects in sensor data quickly and accurately.
These days, they’re using their AI modeling system to gather raw data from LiDAR, cameras, and other sensors in an autonomous car, and model it locally using processors onboard the vehicle. That model is, in turn, uploaded to a central data repository for distribution back to autonomous vehicles in the field. If you’re a science-fiction fan, think of it as a hive-mind.
This system has a number of benefits, including “vastly more actionable information than today’s mapping systems” and maps of a much smaller size that require “a fraction of the data storage and transmission for existing technology.”
Civil Maps’ biggest coup, however, is that their technology allows for real-time crowdsourcing of maps. This is a “major improvement over the lengthy processes that require human annotation.” Using their solution, Civil Maps can offer a constantly updated map that is precise enough to give autonomous vehicles the ability to operate safely.
Any technology like this is only useful if it finds a place in a large number of vehicles. This is why funding from Ford Motor Company is undoubtedly good news for Civil Maps. What’s more, the company explains that they plan to “use the seed investment to accelerate product development and deployment with a number of leading automotive companies and technology partners.”
“Autonomous vehicles require a totally new kind of map,” said Civil Maps CEO Sravan Puttagunta, in a prepared statement. “Civil Maps’ scalable map generation process enables fully autonomous vehicles to drive like humans do – identifying on-road and off-road features even when they might be missing, deteriorated or hidden from view and letting a car know what it can expect along its route. We are honored to work with Ford and the rest of our investor team to pave the way for fully autonomous vehicles at continental scale.”
