Proponents of autonomous driving technology believe that it will bring about a revolution in road traffic – there will be fewer traffic accidents and less congestion; for users without cars (not collecting single and double license plates), it means more Convenient travel. Some of the inherent beliefs about traffic travel are also likely to change.

They are already on the American road: Tesla is already equipped with an automatic driving mode; Uber is running a self-driving taxi in Pittsburgh and is also testing a self-driving truck for long-haul transportation; it is still running in San Francisco these two days. The autopilot Volvo XC90 was stopped by the California government without permission. The vans that Google Waymo and Chrysler worked with were also wandering around. Of course, in these cases, there are human drivers in the car - it is reasonable to say that they should hold the steering wheel and pay attention to the road.

Various traditional car manufacturers - Toyota, Subaru, Volkswagen, BMW, etc., are also adding more autonomous driving or intelligent driving assistance functions to new models, such as automatic braking and automatic steering. Lei Feng.com, they plan to launch fully automatic driving products around 2021.

SebasTIan Thrun, founder and former head of Google's Autopilot program, said: "I think autopilot is unstoppable." But maybe it's not Google's sale of the Waymo van, because Thrun also predicts that the number of self-driving cars is more than traditional. It takes at least 15 years for the car. Some people think that it will take longer. Experts are measuring the main technical barriers to achieving autonomous driving and studying how to overcome them. Science news summarizes five major technical challenges:

Perceive the surrounding environment

For computers, the highway with a clear weather is completely different from the road at dusk and haze. Autonomous vehicles must be able to detect road properties in any weather and under any circumstances, such as strong light. MIT robot expert John Leonard commented: “I have seen promising automatic driving techniques in the rain, but snowing weather (and snow) is very difficult.” This shows that we are still some distance from the real all-weather autopilot.

For self-driving cars, on-board sensors must be reliable, compact and inexpensive; and they must have a sufficiently detailed electronic map to match them so that the vehicle can effectively identify and judge the surrounding environment. Lei Feng.com knows that map data is still incomplete in many countries, which requires time and cost to overcome.

John Leonard is working with Toyota to help the car respond safely to a variety of situations. Many other experts use the data collected by car cameras to create up-to-date maps. In this regard, Leonard said: "Modern algorithms run on data, and data is their fuel."

2. "Unexpected encounter"

A correct response to a particular situation is a challenge – such as a traffic police sign at a red light. Missy Cummings, head of the Duke University Robotics Lab, believes that simple rules based on traffic rules may not always work because it is impossible to write all possible conditions into the code beforehand.

Body language and other environmental clues can help humans understand and deal with these situations. But it's challenging for autonomous cars – like a kid who is about to rush to the road. Cummings said: "Automobiles must learn to extract information, which is the essence of AI."

Her team is exploring a new solution: whether the next step of the pedestrian vehicle can be warned by the display on the body. However, the survey results show that pedestrians tend to ignore these novelty things, and still only judge the road safety by the traditional method of observing the speed of the vehicle.

3. Human-computer interaction

Even if a fully-automobile car is already on the horizon, most "auto-driving cars" will still only be semi-automatic in the foreseeable future. It is a very difficult technical task to divide the responsibility of the driver and the car. How does the car inform the sleeping driver to take over in a situation where it cannot handle it? How does the car confirm that the driver is ready?

In this regard, Chris Janssen, a cognitive scholar at Utrecht University in the Netherlands, said: "In a sense, the driver is still focusing on certain aspects of driving, although he is not driving." His laboratory is studying : In such a situation, how do people assign their attention? They use EEG machines to explore how people's brains react to alarm sounds in both driving and riding situations. Janssen is also interested in the best time to communicate the order and how detailed the order is.

4. Ethical dilemmas

In exploring the ethical issues of autonomous vehicles, MIT cognitive scientist Iyad Rahwan emphasized the problems caused by selfishness:

“People who buy autonomous vehicles want the car to put the passengers (self) first. But they want other people’s cars to put pedestrians (self) first.”

In an online test called Moral Machine, participants need to choose who to save in different situations (classic mother-in-law problem). If the pedestrian is a grandmother, would this be an issue for automatic driving (protection priority)? What if she does not follow the traffic regulations? Society as a whole must decide what principles and systems to follow for autonomous driving. In order for technology to keep up with ethics, these decisions must include moral judgments while avoiding consumer bias against autonomous driving.

5. Network security

In 2015, hackers suddenly stopped a jeep driving on the St. Louis highway – its brakes and steering wheel were remotely controlled by hackers through the car entertainment system. The demonstration proves that even traditional cars have security holes. If used by someone with a heart, it may cause an accident. There is a greater risk of auto-driving cars that get system updates and map data through the cloud. Sean Smith, a computer scientist at Dartmouth College, said: "The more information processing technology penetrates into everyday objects, the more difficult it is to monitor their security breaches."

In addition to terrorists and murderers, Sean Smith can think of other situations in which the car is maliciously controlled: for example, hackers deprive a user of the control of the car and extort ransom.

These are the five major challenges that commercialization of autonomous driving technologies needs to overcome. They don't seem to be complicated, but solving every challenge requires multidisciplinary research progress, as well as advances in the way the industry produces (such as lowering sensor prices), as well as communication and cognitive improvement (ethics) across society.

Dash cam for Honda

Dash Cam For Honda,Wireless Dual Dash Cam,Dash Cam Front Rear And Inside,Honda Crv Dash Cam

SHENZHEN ROSOTO TECHNOLOGY CO., LTD. , https://www.rdtkdashcam.com