Tesla's World's First "Self-Driving" Car
In recent years, artificial intelligence (AI) has been gaining popularity, replacing humans in various industries, and even Go's accomplishments have surpassed humans. Brother K Ji felt that sooner or later the machine would rule the human rhythm, so he began to worry about the singularity of AI outbreak, and even when the YY man-machine war began...
(We are talking about Tesla's death !)
Well, K machine brother thinks that Tesla is a typical representative of the current fusion of technology and cars. The thought of high technology is naturally connected to intelligent identification and judgment, which is inseparable from intelligent computing. The purpose of intelligent computing is to replace people, but the question is, can "they" really replace us? This topic has once again been pushed to the forefront with the emergence of tesla's first self-driving car accident.
The owner is a 40-year-old brother named Joshua Brown. He was once a member of the US Navy SEALs and often showed pleasant pictures of his hands off the steering wheel on social platforms.
Put yourself in the number one position for Write For Us Technology category by sending email with your blog at deskgrass@gmail.com
At the time of the incident, Joshua Brown was comfortable playing a game on the Model S with autopilot mode enabled (another media report said that he was watching "Harry Potter"), and then hit a turning white tow truck at full speed, and No more ...
This incident quickly became the focus of global news. K machine brother once thought that he should focus on Tesla, but later found that the momentum was wrong. A large number of reports believed that this was the first major accident after the Tesla autonomous driving system travelled 210 million kilometres. The accident probability is far lower than that of human driving.
This is somewhat of a self-driving spirit, suspecting Tesla of whitewashing. Compared with the data of about 1.24 million people who died of traffic accidents every year in the world statistics last year, this example of Tesla looks like a nuisance, and the safety and reliability of intelligent driving seem to make sense. But with Brother K's careful analysis of the cause of this accident, I just want to say that this kind of spirituality is completely "Muppet Schuette"! Tesla paid for the "Water Army".
OK nonsense, to understand the cause of this accident, let's take a look at what the Tesla Autoplot is. Too complicated to say, K machine brother simply understood it as the three main parts shown in the following figure.
Tesla promotes that the recognition rate of this system is very good. The three modules work independently and work together. Even if one module affects, it can avoid collisions, scratches, and other accidents ... In short, the X is explosive. So how could this major accident that caused death?
Official propaganda is missing inspections, all three systems are missing inspections at the same time! !!
Okay, let ’s see how the missed detection happens with a very low probability as described by the official. Brother K has sorted it out from the perspective of three systems!
You can read the guideline for write for us fashion categories and contact us at aclassblogs@gmail.com
1. First is Tesla Autopilot's MobileyeEyeQ3 vision system
1. The camera selected by Tesla Autopilot is a telephoto lens, and many times you cannot see the full picture of the obstacle. For example, when the white trailer truck enters the visual area in this incident, the camera can only see the middle of the truck suspended on the ground, but cannot see the entire vehicle.
2. Mobileye engineers have stated that the EyeQ3 system has many difficult cases (Corner Case), and floating objects are one of the more difficult problems to solve. Such as road rescue vehicles and double-decker vehicle transport trucks, Mobileye has difficulty understanding objects floating in the air.
3. Inefficient detection of large white obstacles. Mobileye's monocular camera has experience in detecting distance and speed when performing obstacle detection. However, for obstacle detection of large-area white objects, neither the latest algorithms in academia nor Mobileye can solve the problem of low recognition rate. Because large-area white objects are difficult to extract feature points from images, There is not enough data input to solve the camera pose of the point, resulting in a very high rate of missed detection of large-area white obstacles.
OK, we can easily find three major keywords: incomplete observation, large area of white, and off the ground! Well, this is not the driver who can't see when he is drunk, or the driver who can't see when he sleeps, but sees it clearly and hits it. This means that all large white vehicles with high chassis or obstacles to the ground may be Tesla stealth killers.
2. Next is the Millimeter Wave Radar System
The reason for this failure is simpler and funnier, mainly because its installation position is too low. Generally, the vertical angle of view of the millimetre-wave radar is within 5 meters. The Tesla's millimetre-wave radar is installed near the site of the vehicle's head. As a result, when Tesla approaches the side of the trailer truck, the radar beam passes through the truck from the underside, so it misses detection.
3. Finally, the Ultrasonic Sensor
Sorry, the measuring distance of this gadget is too short (about 5-10 meters), it is useless at high speeds, and it cannot detect longitudinal obstacles. It may be barely possible to stop at a low speed.
Objectively speaking, the improved machine is more reliable than humans and safer to drive. Problems with driving are mainly caused by inattention, drinking and affecting judgement, and the machine will not be tired. But the machine is not perfect at this stage. Brother Joshua used his life to check out a bug in Tesla. Brother K is wondering, how much more life is needed to verify it? This is probably not just a legal issue.
Finally, remind the majority of Tesla owners that your car ’s beauty is explosive, and the sense of technology is bursting, but when you install X in the morning
and prepare to take a photo of going back to work on the way to work, do not really fall asleep, and always pay attention to whether there is white Kcal. Software companies in Pakistan and China are working on Artificial Intelligene too and hope soon the project like this will be use for Public in Asia.
