Expanding Computer Vision and Automation
CHRIS QUIRK
n the summer of 1995, Navlab 5, a Pontiac minivan with a laptop powered by the cigarette lighter, pulled out of Pittsburgh on a 3,000-mile autonomous journey. The No Hands Across America trip made headlines, and since that time, Carnegie Mellon University has been renowned for its computer vision and autonomous transportation technologies.
Thirty years on, School of Computer Science researchers have performed foundational research and design that undergird the industry, with many CMU-related startups across myriad autonomous driving sectors and related industries to show for their efforts. In addition to a wealth of scientific advancement produced through faculty and alumni, SCS spinoffs are shaping the contours of the future in budding fields.
Chris Urmson (SCS 2005), CEO of Aurora
Chris Urmson (SCS 2005) is the CEO of Aurora, an autonomous-driving truck company. Urmson, who served as chief technology officer for self-driving cars at Google for seven years, co-founded Aurora with Sterling Anderson and Drew Bagnell in 2017. The company began the first-ever long-haul driverless commercial truck runs in April 2025. The semis travel between Dallas and Houston, and the company expects to expand to Fort Worth, El Paso and Phoenix routes by the end of 2025.
Urmson pivoted to trucking after examining the market and the challenges of moving freight. “From day one, our goal at Aurora was to build a self-driving system, the Aurora Driver, to power any vehicle — from light passenger cars to Class 8 trucks. Initially, we were focused on autonomous passenger vehicles,” he said. “In 2019, one of my co-founders made a compelling case that we should focus first on trucking, and he was right.” The U.S. trucking industry is a trillion-dollar market, and it faces a growing shortage of human drivers. Trucking also remains one of the most dangerous professions in the country, with on-the-job fatality rates 10 times higher than the national average. “The business and societal case for tackling trucking first was crystal clear,” said Urmson.
Urmson also pointed out that trucks move about 70% of the nation’s goods. “There is a massive opportunity to make transporting freight safer, more reliable and more efficient,” he said.
Getting a big rig to drive itself is a challenging endeavor. After all, they can weigh up to 80,000 pounds and travel at highway speeds. “One of the big challenges we had to solve was perception, how to see far enough down the road to detect potential hazards with enough time to respond safely,” Urmson said. To improve on the available technology, Aurora developed a proprietary long-range lidar called FirstLight early on. “We recognized that no off-the-shelf lidar system offered the range we needed for safe autonomous highway driving," explained Urmson. FirstLight allows Aurora’s trucks to detect and track objects more than 400 meters away, far beyond what other lidar sensors can see, and beyond what human drivers can perceive at night. “Ultimately, it’s important for us to build trust — with the public, with regulators and with our customers,” said Urmson.
Other enterprising startups abound in autonomous driving and essential related fields. Nuro is one example. Dave Ferguson (SCS 2004, 2006) co-founded Nuro, which has licensed its automated driving system and is now operating in California and Texas. Ferguson was a principal engineer on Google’s self-driving program before it became Waymo and was also part of the CMU team that won the 2007 DARPA Urban Grand Challenge.
The Nuro Driver, one of the only driverless autonomous technologies on roads today, making a delivery.
Beyond the driver’s seat, Stephen Smith, a research professor at the Robotics Institute, and Gregory Barlow (SCS 2011), now at planning and design firm Kimley-Horn, created Rapid Flow Technologies in 2015 to improve traffic control. Rapid Flow was a mechanism for Surtec, the traffic technology tool the duo built that uses advanced sensing and AI for adaptive traffic signal control. It increases the efficiency of traffic movement by optimizing signal timings each second. The company was acquired by Miovision in 2022.
Precise mapping is critical for builders and engineers in planning, construction and infrastructure upgrades. Alexander Baikovitz (SCS 2019, 2021), Haowen Shi (SCS 2019, 2021), Zachary Sussman (SCS 2020) and Michael Mong (ENG 2020) have started a company called Mach 9 to promote Digital Surveyor, their geospatial software that creates detailed engineering maps from lidar data faster and at lower costs than existing methods. The 2D and 3D models include details down to the location of light poles and other road features. The company is garnering high-status attention and recently secured funding from Y Combinator, the well-known venture capital group.
Founders of Mach 9 (from left to right): Alexander Baikovitz (SCS 2019, 2021), Haowen Shi (SCS 2019, 2021), Zachary Sussman (SCS 2020) and Michael Mong (ENG 2020)
Agility in the marketplace is a signature feature of many CMU startups. A good idea isn’t always enough, and often business leaders find better niches for their offerings in unanticipated spaces.
One case in point is Hellbender, an AI vision and smart devices company. The company makes battery-powered vision systems that do their processing on the edge, rather than using cloud servers, explains Roger Nasci (SCS 2015), who helped found Hellbender in 2021.
To understand the benefits of edge computing, Nasci gives the example of a camera mounted on a bus in a municipal transit system. At present, these cameras are typically passive and only useful for evidence in the case of an incident. With edge-capable technology, if the bus is too crowded or someone is injured, the device could contact the dispatcher or first responders. “You could have live feedback that would send alerts actively,” he said. “We’re getting into a groundbreaking space.”
What Nasci did not expect out of the gate was that Hellbender would become a manufacturer of devices, but the opportunity presented itself and the company pivoted to meet the demand. “We thought we were finding a place in the market for our sensors, but companies wanted help developing the infrastructure and they were willing to pay for it.” As a result, Hellbender will soon offer a suite of monitoring and other bespoke computer vision devices for healthcare, traffic-monitoring and other uses. “There’s a lot of overlap, and pieces we can take and use between different products,” he said. “We don’t necessarily need a completely new design for every solution in a different space.”
Also based in Pittsburgh, Michael Wagner (SCS 1998, 2002) co-founded Edge Case, a company that entered the field of safety innovations for autonomous driving early and has expanded its client base to companies in agriculture, mining and the military.
Wagner became interested in autonomous vehicle safety while at the National Robotics Engineering Center (NREC), developing commercial systems for companies like Caterpillar and John Deere. “I realized it was about more than technology. You could make a business proposition to entice and incentivize developers of all kinds of AI-driven systems to do the needed work to make them safe.” Wagner said. “That has been very useful for self-driving cars and trucks.
Wagner said that an important aspect of Edge Case is that the safety features are integrated into the autonomous driving algorithms that run the vehicle — they aren’t add-ons. Cars in the fleet send data to the operating system for constant improvements. “If you have a self-driving car fleet, we’re going to tell you that maybe your AI isn’t able to detect a certain traffic situation well enough, and you should probably go back and improve that before someone gets hurt,” Wagner said. “You’re staying ahead of risk.”
One of the more intriguing startups in autonomy, and without a doubt among the fastest, is Boom Supersonic, a company that has designed and built a jet that travels faster than the speed of sound — but without creating the explosive sonic blasts prohibited over populated areas.
Blake Scholl (SCS 2001), CEO of Boom Supersonic
Blake Scholl (SCS 2001), CEO of Boom Supersonic and a veteran of Amazon and Groupon, has a pilot’s license but had never worked at an aviation company, much less founded one. After seeing a retired Concorde supersonic jet in Seattle, he was determined to fly in one. “I put a Google Alert on supersonic jets, and wanted to be the first to know when I could buy a ticket and break the sound barrier,” he recalled. Scholl had the alert up for nearly a decade. “It was just crickets. There was no credible effort to pick up where the Concorde left off.”
In 2014, instead of waiting around any longer, Scholl took matters into his own hands and founded Boom Supersonic, a commercial supersonic airline. “I wanted to work on something I loved, and organized my startup ideas by how happy I would be if I was working on it,” said Scholl. “I thought there would be a good reason why no one was doing supersonic transport, and I’d confidently move to the next thing.” But Scholl found there was no good reason. Instead, in researching the business case for a commercial supersonic company, he found what he described as stale, conventional wisdom warning against it. Scholl forged ahead. Boom was originally conceived to follow the path of the Concorde and fly international flights over the ocean, where the sonic booms caused by flying faster than the speed of sound would not disturb anyone. But along the way, Scholl discovered that you could prevent sonic disturbances on the ground by attenuating a supersonic plane’s flight speed based on atmospheric conditions, and updated his business plan to take advantage of the unexpected finding.
Boom flew its first, boom-free supersonic flight with its prototype, the XB-1, in January of 2025. “It was a pretty big deal because there hadn’t been a new civil supersonic airplane since the Concorde in 1969,” said Scholl. Boom is now building the engine for its upcoming commercial airliner, Overture. The new engine, named Symphony, is a custom-built turbofan under construction that will power the airliner to Mach 1.7, and can double the speed of a commercial flight over water. Scholl says Boom will have Symphony ready for testing by the end of this year, and is looking at Overture’s maiden flights in 2028. “We made the engine from scratch, and that turned out to be the key to unlocking the performance we needed to fly those speeds and altitudes.”
Boom has announced 132 orders for Overture from airlines like United, American and Japan Airlines. “Our goal is to be ready for passengers by the end of 2029.” ■
Copilot for Bike Safety
Clark Haynes (SCS 2008) CEO and founder of Velo AI, will soon be releasing a new AI-powered device that will make biking around town safer.
According to the CDC, almost 1,000 cyclists die and about 130,000 are injured in road accidents in the United States annually. Velo AI Chief of Staff Alison Treaster said that 40% of crashes and fatalities involving cyclists are the result of being hit from behind. Often this is because a driver is inattentive or following the cyclist too closely.
Velo AI's Copilot senses danger and lights up when vehicles get too close to bikers.
Copilot sits on the back of a bike and analyzes what’s happening on the road behind a cyclist using its AI predictive capacity. When the device senses danger for the cyclist, like a vehicle approaching too closely, Copilot reacts with flashing patterns of light to get the attention of the motorist, and an audio signal gives a heads-up to the cyclist. The AI in Copilot gives the device a predictive capacity. “You begin to get a sixth sense of what’s going on before it happens,” said Haynes.
Copilot also captures high-resolution video from the bike, for a record of any incident that might occur. Velo AI is currently taking preorders, and will start shipping Copilots in fall of 2025. ■
