This article is produced by NetEase Smart Studio (public number smartman 163). Focus on AI and read the next big era!
[Netease smart news November 9 news] Last year, nearly 37,500 Americans died in traffic accidents. Self-driving cars will reduce the number of traffic accidents (first, they will not drink, send text messages or scream at children sitting in the back seat), but this does not mean that drivers are ready to give up the steering wheel.
A lot of car accidents:
David Groves, a senior policy researcher at RAND Corporation, said: "After inventing the car, a terrible accident occurred because cars often collided." He said: "We have thousands of people who happen to us every year. The car accident is already numb, but the crash of the first self-driving car will be extremely novel." In other words: It is estimated that it will startle people.
and then? Does the public objection make potential innovations infamous or even unattended, like the Sanrishima nuclear energy disaster or the Hindenburg disaster? These are the core issues of a new study published today by Groves and his joint investigator Nidhi Kalra, a robotics expert who can make some decisions on the RAND Center.
The report answers questions raised by self-driving cars, but it is clear that these accidents will still occur. Just look at San Francisco, Arizona's Tempe, Michigan, Boston, Pittsburgh, PA, or Waymo's mysterious former Air Force base in California. However, the large-scale deployment of driverless cars has not yet been completed, and it is difficult for regulators to know when a fully self-driving car will be officially launched.
The rational point of view is: Let the unmanned cars drive on the road because they cause fewer deaths than human drivers. If human driving causes 37,462 people to die each year due to a car accident, and a driverless car will cause 37,461 deaths, let the driverless cars get out of their way. Contradictory point of view: In the driverless car accident, even if there are thousands of people who are “saved†by an undisturbed, drunken self-driving car, the public may die in a self-driving car accident. The engineer may not mind a less perfect robot, but the public may not be so tolerant.
However, it may be more reasonable for public autos to have their own dedicated roads. But when will it wait? Researchers at RAND have adopted an analytical method called "steady decision-making," trying to put forward some rigorous theories on this issue.
Their conclusions sound a bit cliché: Don't let perfection be a good enemy, but it makes sense. They concluded that if the regulatory agency allows less perfect auto-driving cars to go on the road, self-driving cars can save tens or even hundreds of thousands of lives. As Groves said, "Although we cannot predict the future, we find it hard to imagine that waiting for a perfect future will not result in huge opportunity costs."
Hard data
Self-driving cars are obviously not perfect enough. In fact, we know very well how perfect they are. The 43 companies that test self-driving cars in California must submit a “evacuation report†to the public and point out that each human driver will intervene behind the steering wheel of a self-driving car. Last year's report showed that these self-driving cars have become better and better, and the average distance traveled by the Waymo's cars is 5,128 miles in two tests - quite good! In contrast, Mercedes-Benz's performance is not too good. Today, the driving level of self-driving cars is similar to that of lame drivers. "In a self-driving car, you may be safer than a car driven by a 16-year-old or a 90-year-old," researcher Brandon Schoettle told WIRED in August, "but compared to self-driving cars, there is experience. You may feel safer for a car driven by a middle-aged driver.â€
The researchers studied three basic situations. In one of the tests, when the driverless cars were on the road, they were slightly better than normal human drivers, and their safety was increased by about 10%. In another test, when their safety factor increased by 75%, they crashed on the street again. In the third test, AVs reached near perfection, which is 90% higher than that of human drivers. They found that, in general, a mere 10% increase in safety would save more people's lives and be faster, saving about 3,000 people each year.
Coupled with the ability of the driverless car team to continuously improve, once a special accident occurs, all the R&D of unmanned vehicles can learn to avoid this situation, and then the vehicles can be allowed to make their way in the morning instead of being postponed again and again.
But we do not live in a purely utilitarian world, and human beings are not at risk. “Society tolerates a lot of human error on the road,†said Gil Pratt, head of the Toyota Research Institute earlier this year. “After all, we are humans. At the same time, we hope that the performance of the machine can be much better.â€
Digital-based research can ease panic. When driverless cars inevitably lead to death, panic attacks will inevitably follow. (You will not be too concerned with robotic cars saving thousands of lives unless it saves your children.) "One of our hopes is that the presence of such a sample will ease the rebound based on passion." Kalra, a joint investigator at Groves, said.
After all, although the U.S. Department of Transportation stated that 94% of fatal accidents were caused by human error, the public has been telling polling agencies that they are alert to driverless cars. Of the Americans surveyed, 56% told the Pew Research Center they were not even willing to ride a driverless car. Therefore, the federal government is constantly updating the general guidelines for self-driving cars, and Congress is developing a bill to establish a regulatory framework with the support of both parties. This means that the policies formulated today—and the public’s understanding of the realities of risk—will, from a digital perspective, determine the safety of the road ahead. (From: WIRED Compilation: NetEase See Compilation Robot Review: Rain Egg)
Pay attention to NetEase smart public number (smartman163), obtain the latest report of artificial intelligence industry.
AC multi-channels power meter can measure parameters of multiple three phase or single phase power grid such as electrical variables, energy and power quality parameters. They adopt access mode of external current transformer and modular design which is convenient for user to choose different functions to meet different requirements at field.
Ac Multi-Channels Power Meter,Lora Wireless Power Meter,Power Meter With Lora Wirelss,Mulit-Circuit Power Meter Din Type
Jiangsu Sfere Electric Co., Ltd , https://www.elecnova-global.com