Why now?

The idea of AV was presented for the first time in 1939 by Norman Geddes. And despite it has been only developed by universities and motor companies, was never seen as the “future of transportation” for more than 60 years. What has changed in this century then?

(1) Data storage capacity: in 1956, IBM was launching the first hard drive that stored 5 MB, at a value of 10,000 $/MB35.  A quick search at the internet shows us that today it is possible to rent a space on the cloud by 0.007 $/GB/month36.

(2) Network: in 1969 was sent the first message by ARPANET, between the UCLA host (California University, Los Angeles) and SRI host (Stanford Research Institute). In 1981, the number of computers connected to the “internet” was 213. Last year, more than 1,000 million hosts accessed internet by DNS37.

(3) Information processing (CPU): computer costs has been decreasing while its performance increases. In 1961, the IBM unit 1620 was the most evolved one and the cost by GFLOP (operations per second) was around 1.1×10^12 $/GFLOPS; in 2015, the Intel Celeron G1830 costed just about 0.08 $/GFLOPS38.

(4) Bandwidth: the price of bandwidth has been decreasing over the years. In 1998, the price at the USA was around 1,200 $/Mbps; in 2015, price was around 0.63 $/Mbps, that results on an average of 35% price decrease each year39.

We consider that exponential evolution of technology was the biggest impulse for the introduction of AV in our vocabulary. But along this, the funding for AV development by the government, the introduction of new technological companies on the automotive market or events, like DARPA Challenge, that promote visibility to the subject, were also important for the opening of the Pandora box. Do you remember anymore reasons?

(A big thank you to Prof. Luis Bento for his presentation on the subject)

And so it begins…

The replacement of a human being driving a vehicle for a computational system and adequate tools that allow an autonomous driving is not an idea of this century. A small History context allows one to understand the path since the beginning.

The pioneer event for the AV took place in 1939. The AV was presented at the New York World Fair as a projection of a dream. Norman Geddes, sponsored by General Motors, placed the AV as the vehicle that the world would drive in 20 years28. However, the following years had all efforts on the development of war technology by the biggest car manufacturers.

Only in 1977, Tsukuba University in Japan, built, what it may be considered as the first intelligent robotic vehicle29. With cameras and a motherboard it allowed to detect obstacles and follow lines.

In 1983,  Carnegie Mellon University, in the USA, developed a model, the Terregator30, that used a combination of lasers, radars and cameras to move without human interaction. Three years later, the NavLab31, also manufactured by this university, would have been the first AV to carry people aboard at a maximum speed of 32 km/h.

On the same decade (1980), in Europe, the Bundeswehr München university team developed several projects around the autonomous driving. As a result, in 1994, the team presented a modified S-Class Mercedes-Benz called VaMP, that moved autonomously for more than 1000 km, at a speed of 130 km/h28.

ARGO project32, developed between 1997 and 2001, by Parma university, built a prototype (with image interpretation algorithms) that followed lines painted on a road along 2000 km in six days, with an average speed of 90 km/h with 94% of the time completely autonomous29.

In 2004, DARPA, American defense agency, promoted a challenge to accelerate the development of AV. The participants would have to present a AV that could finish the track created by DARPA. None of the participants finished the challenge that year. The year after, 195 teams participated, Stanford  university won and 5 teams finished the challenge: more than 200 km in Nevada without a driver33. This was a turning point.

The dream is old, but the effort duplicated in the last few years and the investment in R&D for this topic lost limits or barriers. The automotive industry has been pressured by other sectors to develop technology to make this real, specially IT companies that are interested in playing a role in this market.

What is the technology?

To accept a fully automated vehicle we need to understand the levels of automation, but when we try to do that we realise that automated vehicles are already a reality and a proven technology. The level 1 of automation is already available in any new family car sold anywhere in the world. In this level we can find the commonly used Cruise Control (CC) that allows us to press a button and program the motor to maintain a certain speed, or go in an automatic way to a set speed value. More examples regarding this level are: Adaptive Cruise Control (ACC), the vehicle follows the front car and detect if it is accelerating or decelerating, or Park Assist (PA), where the steering is controlled by the automated system but the acceleration/deceleration is controlled by the human driver.

It is also a fact that advanced technology concerning the level 2 starts to be used in not only luxury cars. At this level we have the vehicles that are able to park without any human interaction, intelligent parking assist (IPA), and that with an automated system that controls acceleration/deceleration along with car´s movement in congestion areas, traffic jam assist feature (TJA)5. In sum, this level allows a human driver to get his hands off the wheel but still forces the driver to have his eyes on the road. See Tesla Autopilot4.

Level 3 will be the next level, with some technology already in an experimental phase, where the vehicle will perform in an autonomous way but with human supervision and interaction in specific occasions. These vehicles can carry out any decision like a human being would, but in some situations they will require intervention from the driver. Some companies are testing this technology as another stepping stone to reach the next level and this is the point where the interaction between human and machine starts to be discussed. Some accidents are caused by the fact that the human driver disrespected a specific rule, causing the accident, and maybe this will be the turning point of the whole discussion. Now is when we start questioning the actions of the machine, since it is not only assisting a human driver, but making decisions on its own.

At the same time that the technology evolves, so does the sociological discussion, and only this way we will achieve level 4 and 5, where no human interaction will be required. See Waymo2 and CityMobil23 (level 4). Until then, these first levels are real and understand them will allow us to reach to the next one. We hope this blog helps demystify the technology and contribute to the sociological discussion because this will happen, so as soon as we understand it, the better.