Driverless driving is a technology that makes us sad and itchy. This technology is still developing, but we don’t know whether it can truly move from assisted driving to driverless driving. Regarding driverless driving, if we break it down, many technologies are worth discussing. Should this technology go in this direction or that direction?
For example, when it comes to driverless perception systems, Tesla has chosen a different path from other manufacturers. While everyone trusts lidar, Musk believes that the visual system can achieve true driverless driving.
Speaking of Musk, he is definitely a magical person. Some people worship him extremely and even regard him as a god. But some people regard him as a scourge and just sneer at Musk deceiving many ignorant people. This person proposed a plan to immigrate to Mars, and he seemed to be an ambitious entrepreneur. However, this person also speculated in digital currencies, causing many people to lose their money.
Any one-sided evaluation seems inconsistent with Musk, but if Musk is maverick, that is always okay. Ever since he became famous, everything Musk has done has been somewhat surprising. Now that Musk insists on choosing a visual system, we are not surprised. So let's really compare, is Musk's choice correct, or is it the choice of many manufacturers?
The difference between vision and lidar is mainly reflected in three aspects. Performance, security, cost. Performance mainly involves technical matters, which are related to the actual operation experience and whether it can support the subsequent development of autonomous driving in the future; safety can be said to be the most important, and even safety is not enough, consider other Is it still necessary? The cost is more down-to-earth. If the cost is high, the entire vehicle will definitely be expensive. If the cost is low, the pricing of the entire vehicle will be cheaper. You know, smart cars are very likely to replace traditional fuel vehicles in the future. If the price is too expensive, the market competitiveness will definitely be insufficient, and there will be no market value.
In terms of performance, it is difficult to say whether vision or lidar is better. Visual perception, as the name suggests, is seeing with the eyes. Of course, these are not human eyes, but the eyes of artificial intelligence - the camera. This imitates the process of human vision receiving information, then analyzing it by the brain, and then making corresponding decisions. However, artificial intelligence has not yet developed to that point. Although artificial intelligence's learning ability and calculation and analysis capabilities are very powerful, deviations will still occur and affect driving. In the future, if visual perception wants to meet the standards required for driverless driving, it may still require stronger computing power from artificial intelligence.
Lidar is more traditional. In fact, it is based on the radar test environment and then builds a three-dimensional environment model to judge the road conditions. Although this method is not as advanced as visual perception, it sounds more reliable. The only drawback is that there are often some unexpected obstacles on the road, some hinder driving, some do not, and lidar cannot make correct judgments.
Generally speaking, the vision system is more high-end, but in terms of stability, it is still not as reliable as radar.
From a security perspective, visual recognition’s current analysis capabilities are not even that accurate, and security is definitely not that high. Many times, especially on rainy and snowy days, visual perception will be biased, leading to sudden braking, sudden acceleration and other scary operations. This is a great risk to the life safety of car owners. As for lidar, it is slightly better in terms of safety. I don’t know if I will be satisfied in the future, but there is no big problem now. Of course, in rainy and snowy weather, the effect of lidar will be weakened, so you must use it with caution at this time.
Finally, the cost is mainly concentrated on the camera. Relatively speaking, visual perception does not have high requirements for the camera, it is enough, and it mainly relies on intelligence. Lidar is different. It requires high cameras, so the cost is relatively high.
In summary, we have no way of knowing who made the more correct choice among the domestically produced self-driving cars.
But in terms of advantages and disadvantages, the visual perception that Tesla insists on is not very down-to-earth. Especially after Tesla has experienced so many "accidents", people still have doubts about the safety of Tesla. Lidar is relatively safer, but it also requires further research and development by scientific researchers. If Tesla wants to continue to blaze a trail in visual perception, it must further improve its artificial intelligence technology.
It is a good thing that driverless technology can develop in different directions, which is beneficial to the development of science and technology. However, this kind of development should not ignore people's safety. Previously, autonomous driving technology was clearly not mature and could only be used as "assisted driving". However, major manufacturers still had to "exaggerate" to promote autonomous driving, causing many irreparable injuries. loss. We don’t need to say more about Tesla. There have been hundreds of accidents around the world. As for the domestic brand NIO, some time ago, a young and promising entrepreneur also died young because of his reliance on autonomous driving functions.
I hope that car companies can correct their mistakes and develop autonomous driving correctly.