The recent new wave of the epidemic has left many small partners trapped at home, especially those living in Shanghai. After staying at home for more than a month, a good friend of mine in Shanghai finally couldn't help but pour bitter water on me.
"I didn't pay attention to my work before, but during this time I found that my little bunny cub was really too bad to talk about, and I couldn't use chopsticks to eat, I only liked to use spoons."
I'm worried if you'll run out of food and you tell me your son will only use a spoon? Don't forget your son is only 2 years old!
For a moment I couldn't tell if he was showing off or genuinely annoyed. However, out of the restraint and decency of adults, I still forced myself to endure the roar of my heart and seriously comforted him.
"The child is still young, don't care about the spoon or chopsticks, it is a good child to be able to eat by yourself."
Yes, the "white cat and black cat" theory from Grandpa Deng is so useful.
The same truth is not only to comfort friends' parenting troubles, but also to the route differences in the autonomous driving industry - whether it is a "pure vision" solution or a "multi-sensor fusion" solution, it is a good solution to safely achieve higher-level intelligent driving.
After all, in the final analysis, safety is the biggest concern of consumers about self-driving technology.
Nissan Motors may have grasped this consumer demand psychology, and recently, they held a briefing on the latest assisted driving technology aimed at "zero accidents".
Safety is a two-step walk
Takao Asami, senior vice president of global R&D at Nissan Motor, proposed that in order to achieve safely usable autonomous driving, the problem needs to be divided into two steps.
The first is to clarify the definition of security scenarios.
These scenarios range from a single scenario with the highest frequency of daily driving to a highly complex scenario with low frequency but many traffic participants and a large number of decisions.
If all the scenes can be included in a scene library for the automatic driving system to learn machines, and those dangerous and complex scenarios that form accidents can be identified one by one and disposed of, these accident scenarios can be avoided to ensure safety.
This line of thinking is quite a bit of a counter-argument. There are thousands of driving scenarios, if I can identify and avoid all the dangerous scenes, then the rest is naturally a safe scene.
The second is to use automated operations to improve safety performance.
Nissan proposed a concept of "Nissan Safety Barrier". In this concept, Nissan divides the potential stage of danger, the stage of hazard display, the stage of collision possibility, the phase of unavoidable collision, and the stage of collision occurrence according to the distance between the accident risk and the vehicle, and 5 spatial range levels from A to E.
What Nissan has to do is to identify these risk scenarios, use automated operations to control vehicles to avoid risks in the emergency avoidance link, so as to achieve safe automatic driving.
I have new gear
In both steps, environmental perception is a key link that cannot be avoided. Environmental perception is also the focus of nissan's release at this technical communication meeting, which is called "road reality perception" technology every day.
For perception, Nissan chose the route of multi-sensor fusion. In fact, this route choice is indeed more in line with Nissan's robust technical style.
Previously, Nissan released ProPILOT and ProPILOT 2.0.
ProPILOT enables L2-level driver assistance functions such as full-speed ACC adaptive cruise and lane keeping through a single camera and millimeter-wave radar.
ProPILOT 2.0 adds more surround cameras, millimeter wave radar and ultrasonic radar on the basis of the former, realizing the high-speed navigation function.
This time, Nissan added lidar to ProPILOT's sensor array based on the technology of the first two.
Lidar from Luminar is currently the industry's first-class lidar, achieving a detection distance of more than 300 meters, a lateral viewing angle of 120 degrees, a vertical viewing angle of more than 25 degrees, and a resolution of 0.05 degrees.
To sum up the above parameters in a word, this is currently one of the farthest, widest, and thinnest lidar to see.
With the addition of lidar, ProPILOT's perception refinement has gone up a large level. Through the multi-sensor fusion perception of lidar, camera and millimeter-wave radar, "Road Reality Perception" technology can dynamically track the position coordinates of objects around the vehicle in real time, detect obstacles 300 meters away, and accurately detect small changes in the path of the vehicle.
Figurative this system to the real car, is the "little monster of technology" you see below.
It is equipped with 1 lidar, 10 surround view cameras, and 7 radars.
Actual combat is done
So what can this little tech monster armed to the teeth do? Nissan explains its actual capabilities with two actual complex scenarios.
In the first scene, the test car is driving on a road with a serious obstruction of sight on both sides, and a reversing vehicle that suddenly invades the road laterally appears in front of it, the test car recognizes the obstacle and chooses to change lanes to the right to avoid the obstacle in the case of safe right lane. Seeing this, the performance of the test car is quite normal. However, just after completing the lane change, there was a pedestrian "ghost probe" crossing the road in front of him, and at this time, the test car chose to brake directly.
In the second scenario, the test car is following the normal driving of the front car, at this time, a reverse rolling tire appears on the road, and by this time, due to the obstruction of the front car, ProPILOT has not actually recognized the obstacle in front. Subsequently, the front car chose to change lanes to the left, and only then did the obstacle appear in the field of vision of the test car. However, it is worth mentioning that the test car quickly identified the rolling tires and, after completing the judgment of the surrounding lane situation, directly chose to change lanes to the right to avoid obstacles. But before it was over, just after changing the corner, there were transverse intrusion vehicles in front of me. In the face of this scenario, the test car also chose to brake directly.
For many autonomous driving systems now, if it can actively brake in the face of dangerous situations, we will feel as if it is doing a good job. But in real-life scenarios, in many cases, braking does not avoid accidents.
Through the above two live car tests, I can clearly feel the concept of "Nissan safety barrier" mentioned earlier.
For the new ProPILOT system, if the risk is still some distance from the vehicle, zone C in the "Nissan safety barrier" and the previous situation, it will choose to avoid risk by changing lanes. And if the risk is already imminent to Zone D or even Zone E, it will choose an emergency stop.
In addition, the complex scenario mentioned by Nissan is composed of multiple dangerous links, all of which appear immediately after the first risk is avoided. The new ProPILOT system handles these scenarios well, one of which reflects the high real-time nature of Nissan's "road reality perception" technology, and the other also shows the extremely efficient judgment ability of the new system.
As far as these two real car cases are concerned, the emergency disposal performance of the new ProPILOT is indeed stronger than that of many ordinary drivers, which can be called "old drivers", which may be the "improvement of safety performance brought by automation" mentioned by Nissan. Honestly, I was a little bit stunned.
My only question about the system is whether the risk-off capabilities it currently exhibits can only be achieved in preset scenarios, or whether it can learn more scenarios on its own. If it's the former, then Nissan still has a long way to go, but if it's the latter, then I have to blow it up.
And I only now found out that "Nissan Vision 2030" turned out to be a set of combination punches.
In early April, Nissan released all-solid-state battery technology (small partners interested in all-solid-state batteries can click on our previous "Battery All-Solid-State, Nissan "Showdown"" to learn more), and half a month later, Nissan released the latest assisted driving technology.
My tech appetite has been hoisted by Nissan, what will Nissan release next month?
This is the Nissan that fits its Slogan: Technical Nissan, People, Cars, and Life. And I am also looking forward to what the "Vision 2030" in the mouth of the technical nissan will look like.
-END-
(Remember to follow our video number, 7 motors of the Mustang Mach-E 1400)