Ushr, a Livonia, MI-based company making high-definition mapping software for autonomous vehicle applications, announced last week that it has raised $10 million in a Series A round. The investment was led by Forte Ventures, with participation from EnerTech Capital, Emerald Technology Ventures, and GM Ventures.
Ushr was spun out of a 20-year-old venture called GeoDigital, which maps power lines, railways, and engineering structures for the utility and energy sectors. As Ushr’s vice president of operations Chris Thibodeau recalls, GM came to GeoDigital three years ago and asked it to participate in a trial that involved mapping a 20-mile stretch of road. Although primarily working in the energy sector, the company was using a lot of the same mapping tools, such as LiDAR, that are used in autonomous vehicle prototypes.
Of the handful of teams participating in the GM trial, Thibodeau says GeoDigital scored the fewest number of mistakes, such as lane crossing or touching, despite not having a background in self-driving cars. “At the time, we didn’t know what an autonomous vehicle was,” he admits.
Seeing an opportunity on the horizon, GeoDigital eventually spun out Ushr as an independent company. In April, Ushr officially launched with what it calls “the first production HD map software for autonomous driving.”
“Essentially, we created a high-definition software representation of a roadway,” says Brian Radloff, vice president of business development. “It has all the information about lanes and lane markings, shoulders, and the road’s slope, plus information about road objects like signs and traffic lights.”
Ushr’s HD mapping technology works with vehicle sensors and on-road cameras to provide real-world detail. The company’s control algorithms constantly interpret and communicate vital information from sensors to the vehicle control system, which enables more precise steering, directional predictability, and safer vehicle control.
Instead of simply reacting to sensor inputs, Ushr helps the vehicle proactively plan safer routes and anticipate changes in the roadway, Radloff says. Ushr’s customers are automobile manufacturers and major suppliers. “We work with them on integrating our software with theirs,” he adds.
Now that Ushr has mapped the entire controlled-access highway network in the U.S. and Canada with less than four inches deviation, Radloff says the platform provides the most accurate long- and medium-distance sensing systems available.
Ushr also has a second offering: an application programming interface (API), “which utilizes data and gives it to other control modules in the vehicle,” Thibodeau explains. “That’s where we’re unique. An [auto manufacturer] sources our map and then asks us to work with suppliers to integrate the map data into [the manufacturer’s] vehicles.”
The API also allows manufacturers to design their system with lower cost, and could eliminate the need for a map module if Ushr’s technology can be integrated into other control modules, he says. The API is currently being tested with potential customers to gauge interest.
Radloff says some of GeoDigital’s energy-sector investors were eager to get involved in autonomous vehicle development due to its “huge promise and societal impact,” but they also saw the high valuations artificial intelligence companies were receiving and wanted a piece of the action.
Ushr has about 45 employees total across offices in Livonia and Lompoc, CA. Next quarter, the company plans to install its software in production vehicles and further test the capabilities of its API.
The new investment capital will allow Ushr to advance the development of data acquisition technologies and automated map production using artificial intelligence; build a fleet of data collection vehicles; and test in-vehicle software on automotive grade embedded hardware.
“I think we’ll be the first company to launch HD maps in production vehicles,” Thibodeau says. “We’re building out the tools and technology necessary to scale the company and handle a large volume of data. We have to gather data, then we have to process it. The map has to be nearly perfect so the vehicle can make intelligent decisions.”
