China's Ministry of Industrial Information the latest batch of "road motor vehicle production enterprises and product announcement" changes to extend the public notice shows that four Tesla Model Y models to change the information for the record, the new version of the model to remove the radar program, which is the previous practice of the U.S. market Model Y.
In fact, Musk has been repeatedly jumping between the "Tesla Vision" and "Radar" solutions.
At Tesla's Autonomy Day in 2019, Musk took a direct shot at LIDAR: LIDAR is like a bunch of appendages on a person, it's a stupid thing, and anyone who uses it is doomed to fail.
In May 2021, Tesla began removing millimeter wave radar from its vehicles; then last October, it began removing 12 ultrasonic sensors from the Model 3 and Model Y made for North America, Europe, the Middle East and Taiwan. So far, Tesla sensor program from "8 cameras + 1 millimeter wave radar + 12 ultrasonic radar" multi-sensor program reduced to "8 cameras" Tesla vision program.
But then, Musk and "backtracking".
In February this year, according to the Tesla Forum TFF tracked the vehicle change application submitted by Tesla to European regulators, confirming that the latest self-driving hardware HW4.0 is about to be mass-produced on board. It is worth mentioning that in the upcoming HW 4.0 hardware system, Tesla is equipped with a high-resolution millimeter wave radar (possibly high-precision 4D millimeter wave radar).
And then there is now, the new Chinese version of the model will be changed to "Tesla Vision".
Why use Tesla Vision?
The biggest advantage of Tesla vision is the overall cost, Tesla eight camera hardware costs only about $ 200, while the cost of a set of LIDAR in the range of $ 3000-10000.
Specifically, after Tesla removes the millimeter wave radar, the Tesla vision solution requires only eight cameras, while the cost of a single camera is around $30. For the multi-sensors fusion solution, essential is the lidar in it. Although the cost of lidar has been decreasing in recent years, a single one is still around $600-2000, and the entire autonomous driving system generally requires 3-5 lidars, so the total cost is also floating at $3000-10000.
Multi-perceptron fusion compared to Tesla vision cost is greatly increased, the high cost is a key factor to prevent its commercialization on the ground.
In addition, the second advantage of Tesla Vision is the unity and scalability of the system, which does not need to consider the synchronization problem of different perception units in a multi-perceptron solution and the problem of fetching information when there are differences in perception.
Once Tesla Vision acquires the data directly from the camera, it can be directly transferred to the vector space for generation. In the multi-sensor fusion solution, firstly, the transmission synchronization of different sensors needs to be considered, and the transmission bandwidth and delay need to be matched in order to constitute a complete data system; secondly, when there are differences in the perception results of different sensors, for example, when there is a mismatch between the perception information of LIDAR and camera, how to take credit is also an important factor limiting the flexibility of the multi-fusion solution.
The scalability of Tesla vision makes the process of transplanting the whole system to other models and even humanoid robots smoother, and it believes that the launch of Tesla Semi and Tesla Bot will largely accelerate the commercialization of the company's self-driving software in the field of realization.