28 Aug

DeepMap Company Spotlight

At a Glance

On the road, variables such as fog, fading lane lines, and graffiti on a street sign can be challenging for a self-driving car’s sensors. To solve this problem, DeepMap is developing a solution that provides precise, high-definition (HD) maps and real-time localization for autonomous vehicles. While traditional GPS maps can be off by several meters, DeepMap’s technology lets self-driving cars know exactly where in the world they are, down to the centimeter. For example, if a rainstorm obscures a road’s lane markers with a couple inches of murky water, DeepMap’s HD maps enable the car to continue driving safely between the lines. Instead of using separate, specialized survey fleets to create maps, DeepMap’s software integrates directly into its customers’ self-driving fleets. This means that DeepMap’s maps are, in a sense, “living,” as they receive updates to changing road conditions with every mile driven.

In Depth

DeepMap, based in Palo Alto, California, was founded by industry veterans with extensive experience in cloud-based mapping, artificial intelligence, and self-driving technologies. The DeepMap team has developed a unique approach to address the increasing demand for scalable, high-resolution mapping. Other mapping companies use specialized survey fleets to gather data that they then process – in a mostly manual way – to create maps which self-driving cars use as a frame of reference. Initially, these maps are mostly correct. But what if there is suddenly a construction project blocking the exit or a newly installed stop sign? These changes would not be observable on a map created solely by survey vehicles.

DeepMap’s unique solution enables self-driving vehicles to be both map makers and map consumers. To accomplish this, DeepMap has developed advanced software that can utilize the sensors that are already built into autonomous vehicles and can function without additional hardware.

According to Wei Luo, DeepMap’s COO, this is what makes the company stand apart from the competition. Wei explained that, “Wherever self-driving cars go, we collect the data. That data is then crunched in the cloud and served back to the car, providing near real-time data for the cars to use to navigate. The idea is for this map-creating and map-feeding to create a virtuous cycle where self-driving cars are always running with the most up-to-date map data.”

Because DeepMap’s technology is being developed primarily for Level 4/5 autonomy (i.e., fully self-driving cars), you won’t find it in cars available to consumers yet, but stay tuned. DeepMap is already testing its technology with multiple self-driving fleets. DeepMap aims to “accelerate self-driving into reality,” and it seems that they are well on their way. Investors share their vision. DeepMap has raised $35 million from venture powerhouses such as Accel, Andreessen Horowitz, and GSR Ventures.