
The first truly autonomous car was built by Stanford University's Artificial Intelligence Laboratory in 1961. Named the Stanford Cart, it was a modified line-following cart that could navigate a room without human intervention using early computer vision and a tethered computer system. However, if we consider any vehicle with self-guiding capabilities, a strong case can be made for a 1925 Chandler sedan modified by Houdina Radio Control, which was operated via radio waves from a following car.
The challenge in naming a single "first" is defining "autonomous." The 1925 "Linrrican Wonder" was radio-controlled, not independent. The Stanford Cart, while slow (taking up to 15 minutes to cross a room), was a landmark because it made its own decisions based on sensor input. This distinction between remote control and true autonomy is critical. The field then evolved through key projects like Ernst Dickmanns' VaMoRs van in the 1980s, which achieved high-speed autonomous driving on highways, and the U.S. Defense Advanced Research Projects Agency's (DARPA) Grand Challenges in the 2000s, which accelerated modern development.
The following table outlines key early milestones:
| Year | Project/Vehicle | Institution/Company | Key Achievement | Autonomy Level |
|---|---|---|---|---|
| 1925 | Linrrican Wonder | Houdina Radio Control | Radio-controlled car demonstrated in traffic. | Remote-Controlled |
| 1961 | Stanford Cart | Stanford University | Navigated a room using onboard cameras and algorithms. | Early Autonomous |
| 1977 | Tsukuba Mechanical | Tsukuba Lab (Japan) | Achieved speeds of 20 mph by tracking white street markers. | Autonomous |
| 1980s | VaMoRs | Ernst Dickmanns (Bundeswehr University Munich) | Used computer vision for high-speed autonomous driving on unrestricted roads. | Highly Autonomous |
| 1987 | ALVINN | Carnegie Mellon University | A neural network-driven van that learned to steer by watching a human driver. | Autonomous |
| 1995 | No Hands Across America | Carnegie Mellon University (Navlab 5) | Drove 2,797 miles coast-to-coast, 98% autonomously using computer vision. | Autonomous |
So, while the 1925 vehicle was a technological marvel, the Stanford Cart represents the philosophical origin of the self-driving cars we know today, marking the shift from external remote control to internal decision-making.

Forget a single inventor; it was more like a slow-burn relay race. The first big public splash was actually a radio-controlled car from 1925, the "Linrrican Wonder." It weaved through NYC traffic, but a guy in a following car was steering it with a radio transmitter. The real breakthrough for a car that could think for itself came from Stanford in the '60s with their slow but "Cart." It was the seed that everything else grew from.

As an engineer, I look at the "first" through the lens of capability. A radio-controlled car in 1925 was a clever trick, but true autonomy requires perception and decision-making. That crown goes to the Stanford Cart from 1961. It used a simple camera and a massive, off-board computer to identify and navigate around obstacles in its path. It was glacially slow, but it was the first time a vehicle successfully interpreted its environment and acted independently. That fundamental principle still powers every self-driving car today.

My dad, a huge history buff, would point to the 1920s. He told me about an experiment where a car was driven down the street with no one at the wheel, controlled by radio signals from another vehicle. It fascinated people, but it was like a very long-range remote-control car. The real magic started in the 1960s at universities where researchers were teaching machines to see. That’s when cars began to understand the world for themselves, not just follow commands.

If you're thinking of a car that drove itself like today's prototypes, the journey started in the 1980s with a team in Germany. Their van, stuffed with cameras and computers, could reliably drive itself on empty highways. But the spark was earlier. In 1961, researchers at Stanford built a cart that could slowly navigate a room by itself. It was a lab experiment, but it proved the core idea was possible. The "first" depends on how you define the goalpost.


