News on February 21 According to White Swan Finance, a foreign netizen recently posted a video on TikTok that he was driving a Tesla car in an unmanned cemetery, but the vehicle detected pedestrians around him, which was very crippled.
Since then, other netizens have also posted videos of Tesla cars sensing pedestrians on unmanned streets.
Netizens said it was a little scary.
Coincidentally, on February 15th, in Xiamen, Fujian, after a Tesla entered the tunnel, the screen showed a bus passing by, but in fact there were no vehicles on either side. According to the video exposed by the car owner, the car owner was driving in a tunnel, which was very empty.
However, this weird situation continued for more than ten minutes. Later, the image of the bus disappeared after exiting the tunnel, which shocked the owner. The after-sales service indicated that it might be a radar error detection.
Tesla has a sixth sense? Expert: It may be a real ghost
As we all know, Musk hates lidar.
In the early days of autonomous driving, Tesla made a clear decision to bet on computer vision assisted by cameras and radar.
Tesla uses eight surround cameras around its vehicles, supplemented by 12 ultrasonic sensors and a front-facing radar.
One of the characteristics of Tesla's autopilot system is the ability to match the speed of the car with the traffic conditions on the road. It can also detect road signs and traffic lights, and automatically decelerate. At the same time, when it detects a car or obstacle that may collide with the car ahead, it can also step on the brakes accordingly.
No one thought that this advanced system would first detect the existence of "ghosts".
Experts in this field believe that the systems used in such cars include radar, lidar or ultrasonic sensors. They all use the frequencies emitted by these systems, and we cannot perceive them.
Therefore, because supernatural beings emit electromagnetic energy, it cannot be ruled out that Tesla's security system has detected the existence of the soul.
What's more, even this technique can help investigate them.
Regardless, for skeptics, these examples once again illustrate that driver assistance systems and various "autonomous driving" are far from perfect and sometimes make mistakes.
Controversial detection system
In addition to hybrid vehicles, Tesla's innovations have brought us closer and closer to the future.
It is so advanced that even "ghosts" can be detected, which is really chilling.
As an autonomous vehicle, pedestrian detectors are the guarantee that it can drive on the road.
And it has failed more than once.
For example, this paragraph: Model 3 approaches a rotating dummy.
However, the system failed to detect the dummy, and the Model 3 directly hit the dummy and "dismembered" the dummy before stopping.
In another video, Model 3 was tested on a crosswalk and the results were similar. From the display in the cab, we can see that the system detected the dummy pedestrian but failed to slow down in time.
In order to realize autonomous driving as soon as possible, Tesla is also constantly updating its own systems.
New 4D radar, autopilot range is twice the original
In October last year, the Tesla software hacker "green" broke the news and discovered a new feature tip in the Tesla software update-a new radar called "Phoenix".
The Tesla 4D radar is an upgraded version included in the Model 3 series, which can provide high-resolution images in real time through the system to identify, evaluate and respond to various real scenes.
According to rumors, Tesla is seeking to update its radar with new 4D sensor technology obtained from an Israeli start-up company Arbe Robotics.
The company describes its technology as a breakthrough in radar:
Arbe's proprietary baseband processing chip integrates the radar processing unit (RPU) architecture with embedded radar signal processing algorithms, which can convert large amounts of raw data in real time while maintaining low silicon power consumption. Arbe's patent-pending processing chip can manage up to 48 Rx channels and 48 Tx channels in real time, and can generate 30 frames of complete 4D images per second, with an equivalent processing throughput of 3 Tb/sec.
According to specifications, this will almost double the range of Tesla's current radar to 300 meters (984 feet).
They describe their technology as being able to produce complete 4D images, which is exactly what Musk has been talking about about the rewriting of Tesla's autopilot.
"The current image recognition is still derived from isolated pictures. In fact, these pictures are very related in time. Then, when transitioning to 4D, adding a time dimension to the 3D space, it will be like video. This architecture is It’s really needed for fully autonomous driving.”
Earlier, Arbe CEO and co-founder Kobi Marenko also said, "Ultra-high resolution 4D imaging radar is expected to upgrade the radar from auxiliary accessories to a key solution for autonomous driving systems, from redundancy to the core of autonomous driving."
What do you think of this video?