In 1986, a blue Chevy pickup truck often drove the streets of Pittsburgh, Pennsylvania near Carnegie Mellon University. To the casual observer, nothing about it seemed out of the ordinary. Most people drove by without noticing the camcorder sticking out of its roof or the fact that there were no hands on the wheel.
But if a passerby had stopped to inspect the van and look inside, they would have realized it was no ordinary car. It was the world’s first self-driving automobile: a pioneering work of computing and engineering somehow built in a world where fax machines were still the predominant way to send documents , and most telephones still had cords. But despite being stuck in a time when technology had yet to catch up with humanity’s imagination, the pickup truck – and the researchers crammed into it – helped lay the groundwork for all the prototype Teslas, Waymos and Self-driving Ubers cruising around our streets in 2022.
How the first autonomous car was born
The aforementioned van was designed and built by Carnegie Mellon’s Navigation Lab (Navlab) – long before the World Wide Web or Google existed, and with computers 10 times less powerful than the first-generation Apple Watch.
With funding from the US Department of Defense, the robotics division of Carnegie Mellon created the Navlab in 1984 to explore autonomous navigation. The goal, said Dr. Chuck Thorpe, the computer science professor who led the project at Digital Trends, was to deal with “boring, dirty and dangerous” situations.
The Department of Defense, specifically, was looking to build autonomous scouts. These scouts would go into the field and map uncharted territory, where there was usually a greater risk of mines and hidden enemies – work for which humans had previously risked their lives. And so, the Terragator was born in 1983.
The six-wheeled Terregator, which at first glance could easily be mistaken for the Mars Rover’s predecessor, was the world’s first autonomous outdoor driving robot, and at a time when cell phones weighed 11 pounds, it was a remarkable technical feat. . It featured an array of sensors and computer vision technology to avoid obstacles, climb rough terrain, follow paths and more. Work on the Terregator helped researchers realize the technology’s potential, and three years later the Navlab 1 – that blue Chevrolet pickup truck – hit the streets.
The Navlab 1 was as primitive as a self-driving car could get. It didn’t have the sleek touchscreens or smartphone controls found in self-driving vehicles these days. What he had was half a dozen refrigerator-sized racks of computer equipment, a full-size camcorder sticking out from above the windshield, a 20-kilowatt generator, and a few bulk monitors used for show the algorithm’s performance to a handful of grad students crammed into the back. The whole setup looked more like an FBI surveillance van than a self-driving project.
The way Navlab 1 headed was pretty straightforward. Its lidar sensor – similar to that found on the latest iPhones – would fire lasers at objects to determine its distance from them. On top of that, with computer vision, it would break down video camera footage to track lane markings and determine the edges of the road so it doesn’t swerve. The results of these data points would ultimately help him send the final steering commands.
If that sounds like a lot of work for 1980s computers, that’s because it was. Since the hardware had not yet caught up to these advances, the calculations would take years to produce and as a result Navlab 1’s speed was limited to 20 mph.
In addition, the piles of equipment crammed into the back of the van suffered from limited ventilation and as a result it often broke down and even caught fire once, according to Dr. Dean Pomerleu, who joined the team. Navlab as a Ph.D. raised.
Learn from past mistakes
While Navlab continued to refine its self-driving modules over the coming years, it was not until 1989 that Dr. Pomerleu taught a camouflage-colored army ambulance Humvee – Navlab 2 – for learn from their mistakes as the group achieved their next breakthrough.
Until 1989, Navlab students hard-coded programs to correct faults in the self-driving car when it encountered unfamiliar situations. On the other hand, Dr. Pomerleu’s ALVINN (short for An Autonomous Land Vehicle in a Neural Neutral) algorithm allowed the vehicle to adapt to scenarios it was not programmed for simply by watching how a human driver would react in this case. This meant that the next time Navlab 2 encountered the same scenario, it would not need human intervention. It’s what unlocked the next generation of self-driving cars, and even in today’s AI-based systems, the clues of ALVINN can be found.
Soon, Navlab 2 was cruising at 55 mph on a 102-mile road trip from Pittsburgh to Erie, Pennsylvania. “It was the first really long trip he had done and it convinced me that one day we would see vehicles capable of driving themselves on public roads,” added Dr Pomerleu.
Since Navlab’s iterations depended on an adaptive neural network and not 3D maps like Google’s self-driving car, they could be dropped in any place they hadn’t seen before and worked quite well. It was what ultimately propelled the Navlab Division’s victory lap: a nearly 3,000-mile road trip across the country from Pittsburgh to San Diego in 1995.
The Navlab 5 steered for more than 98% of the trip, with Dr. Pomerleu and his graduate student, Dr. Todd Jochem, taking turns accelerating and braking. And despite the wide variations in road types and terrain, the pair encountered almost no anomalies and journaled the entire experience throughout the trip, including the day they demonstrated it for the first time. . The show tonight‘s, Jay Leno, in what was one of the first online travel blogs.
“I think if you go back in time and get one of these cars now,” Dr. Jochem, who now drives a Tesla Model S, said in an email exchange with Digital Trends, “you’d be shocked to see how identical it is in some situations to the performance you see on cars that are now commercial. Very proud of that.
Navlab team members went on to found and make significant contributions to today’s leading self-driving projects, such as Uber, Google, Tesla, and more. Yet despite the progress made by industry, Dr. Pomerleu thinks an “AI winter,” a term used in academic circles to describe a period of low funding and growth in a field, could be looming. around the corner for self-driving vehicles and Elon. Musk might be to blame.
While Dr. Pomerlue agrees that Musk has helped advance the era of autonomous driving, his approach to autonomy, which relies far too heavily on camera sensors and ruthless driver safety policies, is concerning. . “Ultimately, over-promise and under-delivery is in my view unconscionable and threatens to contribute to another ‘AV winter,'” he added.
At the time of writing, the US National Highway Traffic Safety Administration has announced that it is investigating Tesla for letting drivers play video games on the dashboard screen while the car is driving. on autopilot.
The work of researchers like Dr. Thorpe has therefore not yet reached its finish line. “Thirty years ago I predicted that I would retire in a self-driving car,” he joked, “I guess I can’t retire yet.”