“Guess what car I got to ‘drive’ today,” I greeted my V-8 engine, Motor-City maniac, lover of super-charged cars teenager (aka Dear Son).
“What, Mobility Momma? Autonomous roller blades or a matronly autonomous-vehicle riding lawn mower, or something more dignified?” replied my automotive wise man.
“Guess.”
Dear Son, who actually is interested and a favorite interlocutor of his transportation-tech-wonky mom: “A golf cart, a cute shuttle, an 18-wheeler blazing down the highway?”
Mobility Mama: “Nope. I cruised in a Cadillac around historic Ft. Meyer with the head of Carnegie Mellon University’s Transportation System.”
Dear Son: “Now that should have some lift. You hit 50 miles per hour? Tell me Momma, what did you learn?”
So here is the story.
The author in the back seat of the Cadillac.
I was lured to Industry Day at Joint Base Myer-Henderson Hall – the Army/Marines base next to Arlington Cemetery in Arlington, Va. – with the promise of a ride on a Cushman 6 Golf Cart retrofitted to be an AV by Gaithersburg, Md.’s Robotics Research.
The demonstration was foiled by a very non-AV safety issue – a leaking brake line. So as a replacement, I got the promise of a future tour of Robotics Research’s Maryland testing facility and a demo ride in an Autonomous Cadillac SRX. I also got an unexpected window into what the AV sees on and along the roadway and how it responds.
I should say that the Cadillac is not driverless. Rather, it is a highly automated level 3 vehicle that requires a human operator to take over operations from time to time. The concept is to develop the smarts of the car to a point that it could be driverless. Raj Rajkumar, the head of the National U.S. DOT Center for Transportation Safety, housed at Carnegie Mellon, explained he would shift periodically to manual operations primarily because of issues with GPS and some object-recognition issues.
So, what was it like? Smoother, faster, more informative than other rides in AVs. The caveat is that the driver has to shift the Cadillac from automated to manual mode frequently. Dr. Rajkumar himself did come down before the event to drive the course in order to teach the car about the road. As areas become mapped by mobile sensors, the learning curve speeds up for the “brain” – or artificial intelligence-driven software – in the vehicles.
The sometimes-twitchy ride also had to do with the vehicle stopping because it misperceived some things as pedestrians. At those points, Dr. Rajkumar would then take back over driving. As I could see from a monitor on the dash and through the windshield, the Cadillac slowed when it detected a pedestrian crossing. It automatically moved forward if there was no one and nothing in the crosswalk.
The car could also detect pedestrians crossing another lane in the intersection and anticipate them crossing in front of our car. A small sign in the median that called out the pedestrian crosswalk tripped up the object-recognition system. The Caddie saw that sign and, “thinking” it might be a human in the middle of the road, refused to go any closer in autonomous mode.
During this whole ride, I was seated behind the driver in the backseat. An advantage of this seat is that I had my own plasma screen that displayed what the car saw. Dr. Rajkumar gave us a bit of legend for the screen; curbs, trucks, buildings, other cars in traffic all showed up as different-colored lines and shapes.
Except for the crosswalks and the guide path depicted on the screen, the other objects looked like abstract squiggles. It was interesting to follow the squiggles as we moved around the base and the car adjusted its position and speed. It’s a lot like looking in a rear-facing camera found on many cars and trucks already.
Though there were plenty of pedestrians crossing streets and walking on sidewalks, I saw no bicycles so I did not get to observe how the Cadillac would respond.
Dear Son: “What is your final analysis?”
Mobility Momma: “The Cadillac seems high end both in the ride and in ability to avoid pedestrians. False stops seem likely to be a problem. If I were walking across the street and this Cadillac were bearing down on me, I would be confident that it would stop. And the experience gave credence to concerns that automated vehicles could be stopped on a whim. That might not be such a problem on a university campus or a base like Ft. Meyer-Henderson Hall. Cities and counties still need to figure out how to manage the environment so that AVs, pedestrians, and bikers can peacefully co-exist.”
This is a special addition to my series on how bicyclists, pedestrians, and autonomous vehicles can cohabitate. Part 1 answered Why it is so difficult for AVs to see bicycles? Part 2 suggested Five ways cities and counties can make sure AVs and bikes mix safely.
Upcoming posts in the series will look at how AV-related technology can be used to make pedestrians and bikers safer well before we see lots of automated vehicles on the road and will feature an interview with Uri Tamir, Mobileye’s senior director for strategic initiatives.
Video by Joint Base Myer-Henderson Hall and Carnegie Mellon University on Facebook. Photo by Carnegie Mellon University.