TigerTalks in the City: Self-Driving Vehicles

Dec. 13, 2018

At TigerTalks in the City in early December, driverless cars and the entrepreneurship involved was the fast-moving topic of the evening. Co-sponsored by the Princeton Operations Research & Financial Engineering (ORFE) department and hosted at the New York City headquarters of Yext, this latest edition of TigerTalks was centered on the policies and technologies around autonomous vehicles.

Margaret Holen *95, investor, mentor, and lecturer in ORFE, moderated the discussion. She was joined by Ro Gupta '00, CEO of CARMERA, the street-level intelligence platform specializing in high-definition mapping for autonomous vehicles (AV), as well as Prof. Alain Kornhauser *71, Director of the Transportation Program and Faculty Chair of Princeton Autonomous Vehicle Engineering (PAVE).  

Holen: There are certainly more headlines from the large incumbents, the amount of venture investment in the autonomous vehicles space is $5 billion year-to-date. GM Cruze alone has had $1.5 billion. The amount of investment in large incumbents dwarfs that of startups, so I’m interested in your perspective on whether this is going to be a seedbed for entrepreneurial activity with startups, or is it just going to get harder from here?   

Gupta: I think we’ll continue to see venture activity in this space, because it’s just so multifaceted. In mobility there’s scooters and robo-taxis and autopilot, and so many problems to work on.  

Kornhauser: Adam Jonas from Morgan Stanley pegged the value of what we’re talking about at $10 trillion per year worldwide. Pretty big number. We are at “Houston, we have liftoff!” In the engines, the fuel is going through the nozzle, the hold-down latches are like this [moves hands an inch] and we have moved this far [holds thumb and index finger an inch apart] towards this $10 trillion dollar business. It’s going to be about the same from when we went from the horse-drawn carriage to the automobile.  

Photo of Ro Gupta '00

Ro Gupta '00 discusses CARMERA's work in high-definition mapping for autonomous vehicles. (Photo by Wright Seneres)

Gupta also pointed to SpaceX’s mission to Mars as a useful model for entrepreneurship in the AV industry, which is similarly nascent. Even though SpaceX’s goal is to go to Mars, “the key is to do the research and development work to get to Mars, through test runs to the International Space Station, while doing it at a tenth of the cost of traditional companies doing that work.” “In the fundraising process, you have to have a [gesturing high] ‘wow’ enormous opportunity story, and [gesturing low] you also have to have the de-risking story.”

On the policy side, Prof. Kornhauser argued for driverless cars in places like Central Jersey. In areas with low-income housing far from town centers, these areas have a “mobility disadvantage” and therefore, are in great need of “mobility machines”.

“They should not come to Manhattan. Manhattan is a unique entity. You already have more mobility than you can eat. You have a great subway, at least in Manhattan. Out here on these streets, you don’t need (driverless cars). Now, in outer Brooklyn, in outer Bronx, in outer Queens, in Staten Island, in Newark, in Jersey, in Peoria, in the rest of the nation, absolutely.” Prof. Kornhauser continued, “We don’t have a subway in Princeton,” then said wryly, “although I tried to put the Dinky underground.”

Gupta described how high precision mapping is getting integrated in the AV feature set:

Mapping used to be seen as a nice-to-have, or that AI and deep learning would eliminate the need for maps. There’s a theoretical case to be made that that’s true or could be true one day. But the industry, every single driverless car program except Tesla, uses maps – they won’t even deploy without these maps already in place. It’s not perfect analogy but they’re like virtual railroad tracks that have to be laid down first, to provide confidence in what the world looks like before that robot leaves its garage.

The companies doing the real work, a lot of this is unglamorous, it’s chipping away, it’s using a cocktail of solutions to solve the problem. There’s almost never one silver bullet technology that solves it. Maps are good example of this. For this to be working and accepted by society, it’s not even good enough to be 99% reliable. You have to have 99 and a whole bunch of 9’s after that. The reliability has to be so high. It’s a good way to wrap your head around what’s going on in the ecosystem right now. 

The AV industry is in a Gartner hype cycle, a S-curve with a very heightened time in the beginning, everyone’s excited about what could happen. And then inevitably there’s what’s called the trough of disillusionment, where people’s expectations are not reflected by the pace of development of the technology. However, over time the way technology ends up working, it does catch up with expectations and sometimes more so. It’s reassuring that mapping and other pieces of the technology stack, all these parts of the ecosystem, are forming now.

Photo of TigerTalks in the City: Self-Driving Vehicles

Margaret Holen *95, lecturer at ORFE, speaks at TigerTalks in the City

“It is an interesting time certainly, looking at the materials Alain sends,” said Holen. “He has an amazing weekly update on the autonomous vehicles space, as well as a podcast at SmartDrivingCar.com worth checking out.”

More resources

Relevant to this TigerTalks topic, here is an industry report sharing data from a range of perspectives. And you can read this report with this playlist curated by Margaret Holen as a soundtrack.

Watch the entire discussion here:

PEC logo // story ender