History of Self-Driving Cars

Credit: Getty Images.

Oddly enough, the dream of a self-driving automobile goes as far back as the middle ages, centuries prior to the invention of the car. The evidence for this comes from a sketching by Leonardo De Vinci that was meant to be a rough blueprint for a self-propelled cart. Using wound up springs for propulsion, what he had in mind at the time was fairly simplistic relative to the highly advanced navigation systems being developed today.

It was around the early part of the 20th century that a real concerted effort to develop a driverless car that actually worked started to take shape, beginning with the Houdina Radio Control Company’s first public demonstration of a driverless car in 1925. The vehicle, a radio-controlled 1926 Chandler, was guided through traffic on a route along Broadway and Fifth Avenue with signals sent from another car following close behind. A year later, distributor Achen Motor also showcased a remote-controlled car called the “Phantom Auto” on the streets of Milwaukee.

Though the Phantom Auto drew large crowds during its tour of various cities throughout the 20’s and 30’s, the pure spectacle of a vehicle seemingly traveling without a driver amounted to little more than a curious form of entertainment for onlookers. Furthermore, the setup didn’t make life any easier since it still required someone to control the vehicle from a distance.

What was needed was a bold vision of how cars operating autonomously could better serve cities as part of a more efficient, modernized approach to transportation.

Highway of the future

It wasn’t until the World’s Fair in 1939 that a renowned industrialist named Norman Bel Geddes would put forth such a vision.

His exhibit “Futurama” was remarkable not only for its innovative ideas, but also for the realistic depiction of a city of the future. For example, it introduced expressways as a way to link cities and surrounding communities and proposed an automated highway system in which cars moved autonomously, allowing passengers to arrive at their destinations safely and in an expedient manner. As Bel Geddes explained in his book “Magic Motorways: “These cars of 1960 and the highways on which they drive will have in them devices which will correct the faults of human beings as drivers.”

Sure enough, RCA, in collaboration with General Motors and the state of Nebraska, ran with the idea and began working on an automated highway technology modeled after Bel Geddes’ original concept. In 1958, the team unveiled a 400-foot stretch of automated highway outfitted with electronic circuits built into the pavement. The circuits were used to gauge changing road conditions as well as help steer the vehicles traveling along that part of the road. It was successfully tested and in 1960 a second prototype was demonstrated in Princeton, New Jersey.

That year, RCA and its partners were encouraged enough by the technology’s progress that they announced plans to commercialize the technology sometime within the next 15 years.

As part of their involvement in the project, General Motors even developed and promoted a line of experimental cars that were custom built for these smart roads of the future. The frequently advertised Firebird II and Firebird III both featured a futuristic design and a sophisticated guidance system programmed to work in tandem with the highway’s network of electronic circuits.      

So you’re probably asking “whatever became of that?” Well, the short answer is a lack of funds, which happens to be the case often times. Turns out, the federal government didn’t buy into the hype, or at least wasn’t convinced as to put up the $100,000 per mile investment that RCA and GM had requested to make the grander large-scale dream of automated driving a reality. Hence, the project essentially stalled out at that point.

Interestingly enough, around the same time, officials at the United Kingdom’s Transport and Road Research Laboratory began trialing their own driverless car system. The RRL’s guidance technology was somewhat similar to the short-lived automated highway system in that it was both a car and road system. In this case, the researchers paired a Citroen DS retrofitted with electronic sensors with a magnetic rail track that ran beneath the road.

Unfortunately, like its American counterpart, the project was eventually scrapped after the government opted to cease funding. This despite a series of successful tests and a prospective analysis showing that implanting the system would over time increase road capacity 50 percent, reduce accidents by 40 percent and would eventually pay for itself by the end of the century.

A change in direction

The 60’s also saw other notable attempts by researchers to jump start development on an electronic highway system, though it was by now becoming increasingly obvious that any such undertaking would ultimately prove to be too costly. What this meant going forward was that to be feasible any work on autonomous cars would require at least a slight shifting of gears, with more of an emphasis on figuring out ways to making the car smarter rather than the road.

Engineers at Stanford were among the first to build on this renewed approach. It all started in 1960 when a Stanford engineering graduate student named James Adams set upon constructing a remote controlled lunar rover. He initially assembled a four wheeled cart equipped with a video camera to improve navigation and over the years the idea evolve into a much more intelligent vehicle capable of self-navigating past a chair-filled room all on its own.

In 1977, a team at Japan’s Tsukuba Mechanical Engineering Laboratory took the first major step of developing what many consider to be the first stand-alone autonomous vehicle. Rather than relying on external road technology, it was guided with the help of machine vision wherein a computer analyzes the surrounding environment using imagery from built-in cameras.

The prototype was capable of speeds close to 20 miles per hour and was programmed to follow white street markers.

Interest in artificial intelligence as it applied to transportation grew in the 80’s thanks in part to the pioneering work of a German aerospace engineer named Ernst Dickmanns. His initial effort, backed by Mercedes-Benz, resulted in a proof-of-concept capable of driving autonomously at high speeds. This was achieved by outfitting a Mercedes van with cameras and sensors that collected and fed data into a computer program tasked with adjusting the steering wheel, brake and throttle. The VAMORS prototype was tested successfully in 1986 and a year later debuted publically on the autobahn.

Big players and bigger investments

This led to the European research organization EUREKA launching the Prometheus project, the most ambitious endeavor in the field of driverless vehicles. With an investment of 749,000,000 euros, Dickmanns and researchers at Bundeswehr Universität München were able to make several key advances in camera technology, software and computer processing that culminated in two impressive robot vehicles, the VaMP and VITA-2. To showcase the cars’ quick reaction time and precise maneuvering, the researchers had them move through traffic along a 1,000 kilometer stretch of highway near Paris at speeds of up to 130 kilometers an hour.   

Meanwhile in the United States, a number of research institutions embarked on their own exploration into autonomous car technologies. In 1986, investigators over at the Carnegie Mellon Robotics institute experimented with a number of different cars, starting with a Chevrolet panel van code named NavLab 1 that was converted using video equipment, a GPS receiver and a supercomputer. The following year, engineers at Hughes Research Labs showcased an autonomous car capable of traveling off-road.

In 1996, engineering professor Alberto Broggi and his team at the University of Parma initiated the ARGO project to pick up where the Prometheus project left off. This time, the aim was to show that a car can be turned into a fully-autonomous vehicle with minimal modifications and low-cost parts. The prototype they came up with, a Lancia Thema equipped with little more than two simple black-and-white video cameras and a navigational system based on stereoscopic vision algorithms, ended up running surprisingly well as it covered a route of more than 1,200 miles at an average speed of 56 miles per hour.

At the start of the 21st century, the U.S. military, which began getting involved in the development of autonomous vehicle technology during the 80’s, announced the DARPA Grand Challenge, a long distance competition in which $1 million would be awarded to the team of engineers whose vehicle conquers the 150-mile obstacle course. Although none of the vehicles finished the course, the event was considered a success as it helped to spur innovation in the field. The agency also held several more competitions in subsequent years as a way to encourage engineers to further the technology.  

Google enters the race

In 2010, internet giant Google announced that some of its employees had spent the previous year secretly developing and testing a system for a self-driving car in the hopes of finding a solution that would reduce the number of car accidents each year by half. The project was headed by Sebastian Thrun, director of Stanford’s Artificial Intelligence Laboratory, and brought onboard engineers who worked on cars that competed in DARPA’s Challenge events. The goal was to launch a commercial vehicle by the year 2020.       

The team started out with seven prototypes, six Toyota Priuses and an Audi TT, which were souped up with an array of sensors, cameras, lasers, a special radar and GPS technology that allowed them to do much more than just circumnavigate a pre-determined route. The system can detect objects such as people and numerous potential hazards up to hundreds of yards away.  By 2015, Google cars had logged more than 1 million miles without causing an accident, though they were involved in 13 collisions. The first accident for which the car was at fault occurred in 2016.   

Over the course of the currently ongoing project, the company has made several other huge strides. They lobbied for and got legislation passed to make self-driving cars street legal in four states and the District of Columbia, unveiled a 100 percent autonomous model it plans to release in 2020 and are continually opening up testing sites throughout the country. But perhaps more importantly, all this progress has since spurred many of the biggest names in the automotive industry to pour resources into an idea whose time may very well have arrived.