Back in 1989, a driverless Army ambulance dubbed as ALVINN (Autonomous Land Vehicle In a Neural Network), was cruising around Carnegie Mellon University.
Self-driving cars are not recent technological wonders. Apparently, engineers and researchers have been toiling to devise vehicles that can drive without any sort of human intervention for more than three decades. In fact, research on computer-controlled automobile commenced in 1984 at Carnegie Mellon; the first vehicle, called Navlab 1, began in 1986. ALVINN was run as a test vehicle in the 1990s.
It's worth noting that the earliest report centering on automatic cars technology was published in the mid-70s. Meanwhile, the first full automated van made its debut in the early 80s, according to Motherboard.
This driverless vehicle resurfaced on the internet via a Twitter conversation between two engineers: Oliver Cameron and Dean Pomerleau. Cameron heads an Udacity-based, open source self-driving car project, while Pomerleau is a CMU professor who led the self-driving car project that paved way for ALVINN.
— Oliver Cameron (@olivercameron) November 23, 2016
Cameron posted a video clip shared by some of his students, showing a car steering itself unsupervised with the help of just a camera, The Verge reported.
This led to Pomerleau asking a few questions about neural networks and deep learning. Following some Q and A, Pomerleau mentioned ALVINN, which ran an operating system of 100 million floating point operations per second.
ALVINN featured a CPU that was the size of a fridge and drew power from a 5,000-watt generator. Nevertheless, this didn't stop it from hitting 70 mph by early 1990s.
Bearing this in mind, ALVINN can be aptly deemed as "the predecessor of current self-driving cars", according to Cameron.
The technique ALVINN adapted was using a neural network to drive the car, which was nothing short of avant-garde at the time and was rapidly becoming a favored approach when it comes to self-driving cars, Cameron added.