Tesla accident shows that autonomous driving technology is a key breakthrough area of ​​artificial intelligence

The Tesla Model S car was recently involved in an accidental accident, when it was probably operating in auto-navigation mode. The accident sparked a new round of discussions about the most advanced automotive automation technology available today. However, as the investigation into this incident deepens, there is still a very strong debate that automation technology, including artificial intelligence research on autonomous vehicles, can help reduce 94% of fatal crashes caused by human error.

Artificial intelligence is being applied to activities that were previously considered to be exclusive areas of humanity. Most complex human activities can be broken down into several different tasks or steps, some of which are more disciplined in nature and therefore more suitable for automation, while others require judgment, problem-solving capabilities, and others that are difficult to automate. Ability. Automating the more conventional parts of these complex activities allows us to focus on tasks that require more attention.

Tesla Accident: Autopilot technology is the top priority of artificial intelligence research

This is the case of applying automation technology and artificial intelligence to car driving. The NaTIonal Highway Traffic Safety AdministraTIon (NHTSA) defines five different levels of autonomous driving:

No automation (level 0): The driver needs to control the car completely and individually at any time.

Automation of specific functions (level 1): Drivers have absolute power to control the car, but are automatically assisted in certain normal or potentially dangerous situations. For example, electronic stability control and anTI-lock braking systems.

Automated Combined Function (Level 2): ​​The driver transfers the mastership of at least two primary duties, but is still responsible for monitoring and security operations, and is guaranteed to complete the task at any time. For example, adaptive cruise control (adapTIve cruise control) and lane departure warning systems.

Limited autopilot (level 3): Under certain traffic and environmental conditions, the driver can transfer all control, but the driver must be able to take control at any time if necessary.

Fully automatic driving (level 4): Humans provide destination and/or navigation inputs, but it is not necessary to control the car at any time while driving. The car is solely responsible for safe operation, either when it is full or when it is empty.

The automotive industry has achieved Level 1 performance for nearly forty years, and has gradually added more sophisticated Level 2 performance over the past few decades. In general, these properties are first introduced to high-end cars, and then with the cost reduction, low-cost cars have these properties. While some performance is user-friendly, security is always above all else.

According to NHTSA, there were more than 35,000 traffic fatalities in 2015, an increase of 7.7% over 2014. In addition, more than 2.3 million people were injured in a car accident in 2014. 94% of car accidents can be attributed to human error. In the past decade, nearly 100 people have died in traffic accidents in the United States alone.

These data are cruelly warning us that, as NHSTSA manager Mark Rosekind pointed out, "we need to focus on improving human behavior and improving the technology of cars, not only to protect people from car accidents, but also to Preventing car accidents for a while.” In order to significantly reduce 94% of fatal car accidents caused by human error, automotive automation research, including autonomous vehicle intelligence research, is a very important priority.

For example, Toyota Motor recently established the Toyota Research Institute to develop an automation technology called the guardian angel. Under this technical condition, drivers have mastered the power before the car accident. And before the car accident occurs, the automation system takes over completely to avoid the car accident. Toyota is also pursuing the development of a level 4 autonomous car, but its guardian angel technology seems to be more promising in the near future. The technology is clearly positioned as an advanced Level 2 performance, allowing drivers to always control the car unless an accident is quietly approaching. Its key purpose is to protect human drivers who make frequent mistakes instead of replacing them.

In the past decade, nearly 100 people have died in traffic accidents in the United States alone.

While almost all car companies continue to spend huge sums of money to develop advanced Level 1 and 2 technologies, their approach to Level 4 is quite different. Some of these companies, such as Tesla, use a more gradual level 3 path that allows drivers to give up control of the car under certain conditions, while emphasizing that drivers must remain vigilant, committed, and ready to take over. However, as a number of recent articles have pointed out, the autopilot function is so fascinating that drivers will soon become comfortable and relaxed, and mistakenly think they can shift their attention to the surrounding roads. Then they were distracted by other activities and took their hands off the car, regardless of whether they were repeatedly warned not to do so. "The experiments conducted last year by Virginia Tech researchers and supported by the National Security Agency found that drivers of Level 3 cars took an average of 17 seconds to respond to the task takeover request," a recent article in the New York Times wrote, "At that 17 In seconds, a car with 65 miles (104.60736 km) of mileage will travel 1,621 feet (494.0808 meters) - more than five football fields."

According to a recent article in the Los Angeles Times, Google chose a different path than Tesla based on the early experience of autopilot car prototypes. "After the steering wheel of the improved Lexus SUV, the driver can quickly start looking through their bags, banging their phones, no longer having to put their hands on the car. Driving at a speed of 60 mph On the road, the driver can do all of this.” At an expert discussion earlier this year, Chris Urmson, director of Google’s automated driving program, stated, “In about five minutes, everyone thinks the car is working well, then, They will believe that the car will continue to perform well. And the point we want to talk about is that people are doing stupid things in the car."

"After witnessing how people misused technology and ignored warnings about beware, Google chose to modify its algorithms until they were able to protect humans... focus on fully automated cars, that is, driving alone without any communication with people. The car, but currently, this car is only operated under the supervision of Google experts."

However, fully automated cars still face many challenges. For example, Gill Pratt, who was dug up by Toyota from the US Department of Defense's Advanced Research Projects Agency (DARPA) and led the Toyota Institute, pointed out earlier this year that fully automated cars must be able to handle extremely unusual and never encountered before. Situations, such as avoiding mattresses falling from trucks on crowded roads. It takes time to meet these challenges.

Considering that fully automated Level 4 cars are likely to appear many years in the future, some car companies believe that a more gradual transition to Level 3 is still an appropriate step, especially if this step is aimed at reducing The overall traffic accident was completed under this highly specific condition. For example, Audi is planning to introduce Level 3 performance for some of its vehicles in the near future, for example, to deal with traffic on the road at speeds of less than 35 mph.

Finally, there will be other innovations that will go beyond the level of automation defined by NHTSA. While automated machines are difficult to work in highly unpredictable environments, we have been adapting and simplifying the environment so we can benefit from what the machine is good at. In highly discerning environments, such as moving between airport terminals, the shuttle can be fully automated and does not require a human operator.

These methods enable us to design a car that works in harmony with the vehicle's operating environment. We can imagine that cars and trucks can only enter self-driving mode when driving in a self-driving car lane. Once the lane is opened, the vehicle will return to the human driving mode. This hybrid vehicle automation method may be a reasonable intermediate excess before the vehicle is automated.

Research on artificial intelligence in driverless driving is very important because it has too many bets and it can save thousands of lives around the world. Most of the artificial intelligence technologies currently being developed will greatly improve the overall safety of the car and help reduce the number of traffic accidents, deaths and serious injuries. Maybe at some point in the future, we will see autonomous vehicles running on streets and highways.

High Current Inductor

220Uh Inductor,High Current Inductor,Power Inductor,Toroidal Core Inductor

IHUA INDUSTRIES CO.,LTD. , https://www.ihua-transformer.com