A number of failures of Tesla’s autonomous system had been publicly uncovered on TV

The steering wheel and the screen.  The two tools seen inside a Tesla Model 3 with the Autopilot system activated.  Outside, radars, sensors and cameras, watch the environment
The steering wheel and the display screen. The two instruments seen inside a Tesla Model 3 with the Autopilot system activated. Outside, radars, sensors and cameras, watch the surroundings

When you discuss Artificial intelligenceAlways, inevitably, there may be discuss of a person educating the machine every part it should do in every circumstance. That was made clear by the Marcela Riccillo, PhD in Computer Science from the UBA and Expert in Artificial Intelligence and Robotics, when he advised Infobae in August that “We need to be clear that one factor is Artificial Intelligence, and one other very completely different factor is Complete Artificial Intelligence, which was beforehand referred to as Full AI and now has an up to date title that we name Artificial General Intelligence (AGI), and that principally doesn’t exist, as a result of it will imply that machines ought to have conscience, one thing that for now could be solely a part of the science fiction of the films. “

This introduction makes sense to explain a situation that has occurred in the last days with the Tesla Autopilot system, tested by Michael Ballaban, a journalist specialized in Transportation from the American chain CNN, in the transit of one of the largest cities in the world: New York.

When there are no unpredictable situations, the Autopilot system works perfectly, but problems appear when obstacle overlaps occur.
When there are no unpredictable situations, the Autopilot system works perfectly, but problems appear when obstacle overlaps occur.

The aim of the test was to verify how the autonomous driving Autopilot system reacts in heavy and urban traffic that Tesla brings in their vehicles, and that for a few months, it is working in beta with some drivers what were previously qualified based on their background in order to be part of the development of the system.

Development, precisely, is the work that engineers must carry out once the vehicle is equipped with the cameras, radars and sensors what does it require for detect obstacles on their way.

AND the video that aired on the North American network was sufficiently demonstrative of the long road that still has to be traveled to achieve a set-up that allows this autonomous driving safely for the car and especially for its surroundings.

Lane markings may also need to change to be better interpreted by autonomous systems.  The car should have stopped if it had a truck alongside, and not follow the stopped cars straight ahead
Lane markings may also need to change to be better interpreted by autonomous systems. The car should have stopped if it had a truck alongside, and not follow the stopped cars straight ahead

One of the first challenging situations for the Tesla Model 3 that was used at the time, was to be in a two-way avenue circulation with two hands for each side, and with a demarcated island on the floor to turn left. As you approach the junction, but with cars on its right side and cars stopped in front, the system did not stop its march but went straight, with what his driver had to take the wheel to pass between the two. The car must have stopped seeing that the lines on the floor diverted it to the right and that there was a truck on that side.

The second complex test was with a cyclist that, again on a dual circulation avenue, got on Tesla’s radar from the side though in the same direction of traffic. The system Autopilot automatically veered to the left, without noticing that a mail truck was coming up front. Surely, when he was close to the collision, he would have braked, but it would be too late, because he had entered the lane on the wrong side.

The Autopilot system correctly detected the car, but had not seen the metal fence ahead.  Those are the settings that need to be further developed
The Autopilot system correctly detected the car, but had not seen the metal fence ahead. Those are the settings that need to be further developed

The last challenge was when approaching a stopped car on the right hand, through which the driver was traveling. The radar correctly detected the car, but did not see that ahead, about five meters or maybe more, there was a fence demarcating the place where the obstacle was parked. And when he spotted her, he was already too close. The driver-journalist had to take the wheel because if he didn’t, I would have hit the barrier metal in question.

There was also a truck that came in the opposite direction but from its corresponding lane, just when the Tesla wanted to take a corner to turn left, but the truck began to cross the intersection and stopped suddenly because of the traffic, leaving part of its trailer on the crossing itself. ANDhe Autopilot seemed to calculate that it had already gone off the radar and began to turn to your left as scheduled, but again, the human driver intervened to straighten the y direction delay the turn a few meters without impacting the truck.

The Tesla Model 3 was the car of choice for CNN testing in New York, and many of the errors were not from the Autopilot system but from human behavior
The Tesla Model 3 was the car of choice for CNN testing in New York, and many of the errors were not from the Autopilot system but from human behavior

That’s what the development of the Artificial intelligence. To anticipate the possible situations that may arise and program the responses of a system to perform the task correctly. But the world will have to be ordered much more in all respects if you intend to authorize full autonomous vehicles at some point. The Autopilot system requires further development to avoid collisions that shouldn’t happen, but not all problems are from the machine.

The erratic behavior of humans cannot be foreseen by the machine that has no judgment.  Men must learn to coexist with machines so that a world of autonomous vehicles is possible
The erratic behavior of humans cannot be foreseen by the machine that has no judgment. Men must learn to coexist with machines so that a world of autonomous vehicles is possible

The streets have to be more clearly demarcated, without intersections that confuse a machine, and people should act in a predictable way, knowing that are interacting with machines who have no criteria and they do not know how to solve situations that man can cause with unforeseen reactions.

The change in mindset will come one day, but it will probably take longer than imagined. The new generations are born with automated, predictive and tactile systems that will allow them to better understand how to act to be able to share life with autonomous systems. It will not be an easy task.

KEEP READING

A Lamborghini surpasses a million dollars at auction thanks to an autograph
Daytona SP3, the new exclusive Ferrari that only a select few will be able to buy
In depth, how is the perfect decide as much as attain the top of the world

Source link

About Staff

Check Also

Know the precise place the place WhatsApp saves your conversations

Currently, WhatsApp is the moment messaging service most utilized by hundreds of thousands of customers …

Leave a Reply

Your email address will not be published. Required fields are marked *