Use wechat scan QR code < / P > < p > to share with friends and circle of friends < / P > < p > <-- On June 9, < a target = "_ blank" href=" https://news.163.com/news/search?keyword=%E7%89%B9%E6%96%AF%E6%8B%89 "> Tesla < / a > owner Kevin · < a target ="_ blank" href=" https://news.163.com/news/search?keyword=%E5%8F%B2%E5%AF%86%E6%96%AF "> Smith < / a > (Kevin Smith) recently released a < a target ="_ blank" href=" https://news.163.com/news/search?keyword=Model%20Y "> model y < / a > has carried out extreme forward collision warning (FCW) and automatic emergency braking (AEB) stress tests, and the results show that it can successfully identify and avoid obstacles < p > Smith's 2021 model y does not have any radar hardware, but is equipped with what Tesla calls "Tesla vision" system, which consists of eight external cameras. He hopes to evaluate the ability of this all electric crossover vehicle to identify and avoid obstacles on closed roads. Smith said that this is the second round of the test, he used the inflatable clown as an obstacle, model y must avoid this obstacle in order to be considered successful < p > in May this year, Tesla announced that model 3 and model y < a target =, which will be produced in May 2021 and later, will_ blank" href=" https://news.163.com/news/search?keyword=%E6%B1%BD%E8%BD%A6 "> remove radar from vehicle < / a >. The company made plans a few months ago to try to transition its auxiliary driving system to a camera based approach. It selected two Volkswagen models as test objects to collect data for future use. Model s and model x retain the radar, but for a short time. Eventually, the two Tesla flagship models will also switch to the "Tesla vision" system < p > in the short term, model 3 and model y are the only two vehicles that will not use any radar hardware. When Tesla announced that it would remove all radars from the vehicle, many driver assistance systems supported by Tesla's fully automated driving (FSD) software < a target = "_ blank" href=" https://news.163.com/news/search?keyword=Autopilot "> autopilot < / a > owners are worried, and the most common complaint is their performance in bad or low visibility weather. On the contrary, Tesla believes that comparing the visual drive system with the human driver, the vision based method will actually make the vehicle more powerful < p > Elon Musk, chief executive of Tesla, commented on this in the first quarter of 2021 earnings conference call: "when the visual system works properly, it performs better than the best human drivers because it has eight cameras, just like you have eyes on the back of your head, on both sides of your head, and three eyes with different focal lengths looking forward. It's like dealing with situations at superhuman speed. In my opinion, there is no doubt that with pure vision solutions, we can make cars that are much safer than ordinary drivers. " < p > in Smith's test, it is proved that Tesla's camera based driving method definitely needs more improvement. Tesla has released software update 2021.4.18.1 to improve the performance of autopilot based on vision system. The model y can successfully stop when it starts autopilot and encounters obstacles, which is based on the results of several tests on the inflatable model However, in the manual test, Smith controls the car and relies on FCW and AEB to stop the car automatically, but it is not very successful. All manual operations activate FCW, but AEB only works a few times. The car didn't stop either. It just slowed down and reduced the impact. But unfortunately, the inflatable clown is still hit by air leakage < p > other car owners said that compared with other vehicles equipped with cameras and radars at the same time, autopilot performed better with the support of Tesla vision system. Tesla has also added several safety features and new features to its vision only vehicles. Tesla will become more confident because its neural network collects more data on cars equipped with Tesla vision system, which will become more popular in the coming months（ Small) < / P > < p > <-- EndFragment-->