usatoday24
Autonomous driving. That eternal promise that is supposedly worth billions of euros, that we are always brushing with our fingers and that, in the end, never seems to arrive. The last to express their doubts have been those responsible for NVIDIA.
“It’s super difficult”. The words are from Ali Kani, head of NVIDIA’s automotive division, who spoke with the magazine Coach about the future of autonomous driving. “It’s something typical of the next decade. We are not close. It is super difficult,” Kani assured journalists.
The question referred to “truly autonomous cars” of which the head of NVIDIA. It is clear that “we will not see them in this decade.” Kani pointed out that cars currently do not have the power or the technology necessary to implement autonomous driving in the short term.
A new way of working. Regarding how they work at NVIDIA, Kani assures that what they are doing “is very different” from what they did a year ago. He assures that they are “working on large language models such as GPT Chat with video.” And he emphasizes: “no one was doing this three years ago.”
The problems. Basically, we do not have enough computing power to guarantee autonomous driving in the short term. “This type of model needs a lot more computing power, a lot more memory bandwidth. You need more LiDAR sensors and radars, redundant algorithms to make sure it’s safe. All of that has to work in parallel, which means more computing.”
Furthermore, NVIDIA is clear that it is not just a matter of raw power. It’s also about giving a good image to potential customers and showing that you can travel safely. “The industry needs to go slow with this. If one company makes a mistake, the entire industry is set back a few years. So we have to act in the most responsible and don’t take any shortcuts. You can only do it when you have proven that it is really safe,” concludes Kani.
A different approach. If we take what NVIDIA claims as a reference, the company works in a completely different way than Tesla tries to do. Technology makes it clear that radars and LiDAR sensors continue to be essential, components that Tesla wants to eliminate to trust everything to the use of cameras and recorded images.
What Elon Musk’s company defends is that It has a huge fleet of vehicles already on the streets and that everything recorded by them allows their algorithms to learn faster than the competition. They hope, therefore, to spend less money and time to go beyond where Waymo or Cruise have gone.
In 2027. The perspectives that come to us from NVIDIA are also very different from the promises that Elon Musk has made about his Tesla Cybercab. According to the owner of Tesla, his fully autonomous cars should be on the streets in 2027 despite his claims “Be unoptimistic with deadlines”.
Then he said that next year he will be manufacturing his robotaxis. A promise that, as on so many other occasions, doesn’t seem too realistic. However, Musk is confident in his approach to Donald Trump to open the hand with the tests of autonomous vehicles and deploy their services more quickly.
The eternal promise. The truth is, whether we’re talking about Tesla or any other company, the promise of the fully autonomous car always seems to be about to arrive. And it never seems to end. Waymo keeps going but Cruise has fallen by the wayside and has joined a long list of failed attempts.
The truth is that Cruise has managed to launch the service in limited spaces but it is also true that the behavior of its cars it was easy to manipulate and who has lived in a eternal controversy of accidents and encounters with San Francisco emergency services.
During all this time, Tesla has not managed to make its Autopilot a truly autonomous service and requires human attention. Ford’s BlueCruise also needs it, although in this case can be driven without hands on previously mapped roads. Mercedes does not require it but your system is limited to very specific circumstances.
And a bag of millions. Despite everything, there are billions at stake with autonomous driving. Or that’s what they promise us. One of the most optimistic forecasts is that of Tasha Keeney, director of investment analysis at ARK, who quantified in one of his analyzes collected by The Wall Street Journal The value of a robotaxis service like the one Tesla is looking for could account for 60% of the company’s income in 2029 and raise its valuation to $800 billion.
Photo | tesla
GIPHY App Key not set. Please check settings