When will we have fully autonomous vehicles?

It would be almost a utopia to think that we would have them fully functioning within the next decade. But it would be foolish to think that they would not exist within the next 20 to 30 years.

Most people underestimate the technological effort necessary to develop an autonomous vehicle for all kinds of everyday situations.

It is much more difficult than we imagine.

Many advances have been made in terms of algorithms necessary to develop the AI ​​that will allow these vehicles to actually drive themselves.

Computers have an overview of their surroundings, which allows them to know exactly where the vehicle is located, they "know" where each pedestrian or object is on the road, but that is not enough to drive autonomously.

What is really necessary to know is what would happen next. In the next seconds, in the next minute or hour. In other words, systems need to predict the possible events that may arise in the near future. This is the main obstacle encountered by the AI ​​developers of the Massachusetts Institute of Technology in Boston.

Coding algorithms that interpret subtle expressions on the face of a pedestrian or driver is not an easy task. Sometimes a series of gestures or body expressions could indicate if a person is going to cross the street, which would be a trigger for events in the next few seconds.

People have developed an intuition, an instinct product of experience or perhaps because of our very human nature, which allows us in most cases to assume the decision that a pedestrian would make. But these are aspects of great complexity to be precisely programmed.

The fully autonomous car is a distant dream, but intelligent assistance systems are already making driving much safer.

The Tesla company incorporated in its vehicles a autopilot system that also promises to have full self-driving capabilities in the future, through software updates. A series of cameras and ultrasound sensors installed throughout the vehicle have the ability to detect hard and soft objects at a great distance and under adverse weather conditions.

But in the words of Tesla itself:

The autopilot allows your car to drive, accelerate and brake automatically within your lane.
Current autopilot features require active driver supervision and do not make the vehicle autonomous.

Ethical Aspects in Autonomous Driving

Driving a car is not as trivial as you think.

This has to do with the fact that one has to constantly monitor what happens around us: cyclists, pedestrians, animals and other vehicles, weather conditions ... all these factors are binding when making a decision at the wheel.

For some years, the MIT MEDIA LAB in Boston has been dealing with the Ethical factors posed by the development of AI, where the renowned professor Iyad Rahwan and his team developed the so-called "MORAL MACHINE".

They are aware that in most cases people cannot be judged for having made a wrong decision, in situations where they barely have fractions of a second to react.

On the other hand, autonomous driving vehicles have extremely fast cameras, computers and sensors that allow the environment to be detected millions of times per second, making the computer "time goes by slower", allowing you to evaluate the process and then "decide" The best strategy.

As we know, the basis of Artificial Intelligence is the collection, analysis and selection of millions of human responses and reactions. With the use of the "Moral Machine" millions of responses or decisions have been obtained from people around the world.

With a series of 13 hypothetical situations, it is proposed to people who choose whether someone should "die" or "live" according to the possible behavior of the autonomous vehicle.

_{Moral Machine}

When responding to these tests, people are subjected to various ethical, human and cultural factors, which vary greatly depending on the nationality, age and even the gender of the participant.

Who would you save? Would it be the same if you are male or female? If they are children or elderly?
I invite you to try the Moral Machine.

Conclusion

These are part of the ethical and moral dilemmas that AI applied to autonomous driving must face, especially if we think about the fact that errors in this matter will not be allowed, since when the first accident occurs due to a failure in autonomous driving, This will trigger a series of legal events that would delay all the progress that has been achieved so far.

AI offers great opportunities in everyday life, in health and even in "Mobility," but just look at the technical possibilities, we must ask ourselves:

What is the meaning of all this Progress?

"Smart algorithms cannot answer this question. Only human beings can do it."