Sep 7, 2024

Billions of R&D investment in autonomous driving is not as good as taking a Didi?

Clearly, whether or not to collect information in a timely, comprehensive and accurate manner will have the most direct impact on subsequent actions. So this time let's take a look at one of the three heroes of smart driving hardware - the sensor.

Let's talk about these three heroes first. If we consider a smart car as a robot, its actions are designed through the "perception", "decision-making" and "execution" of these three heroes.

Today's common sensors are cameras, ultrasonic radar, millimeter wave radar, and lidar. Those high-end brands also join infrared imaging devices, which is relatively rare for the time being.

The development of automotive electronics is certainly not a one-time thing. Looking back at their history, it's not hard to find that hardware may be small, but it has become a record of the evolution of automotive technology.

The overall timeline of AD started around the year 2000. Around that time, Subaru, Toyota and BMW were among the first to introduce active safety features related to AD. However, at that time, the adopted perception hardware was relatively single, Subaru only equipped with dual cameras, while Toyota used a single millimeter wave radar as the main sensor. The functions achieved are also relatively limited, mainly including lane keeping and ACC etc.

5 years later, more car manufacturers began to enter the active safety field, hardware and functions have been strengthened, with Volvo as a representative. It adopted a single camera + millimeter wave radar fusion scheme, achieving ACC adaptive cruise control, collision warning and lane departure warning, etc. These technologies are not only Volvo's latest safety technology at that time, but also leading in the automotive industry.

Time came to 2010, these five years belonged to the stable development period, although there were not too many revolutionary new technologies appeared, but each brand in orderly display of their own intelligent driving system, sensing hardware, some car companies continued to explore single camera scheme for cost reasons, while others tend towards multi-camera scheme.

Time continues to move forward, with the release of Tesla Autopilot 2.0 in 2016, intelligent driving began to accelerate, and routes that emphasize visual perception with multiple cameras + millimeter wave radar + ultrasonic radar began to appear. We first discovered that side skirts, B pillars could also be equipped with cameras, IMU inertial measurement units would even be installed on the vehicle, of course, the functions achieved were also more abundant, such as automatically following traffic flow and automatic lane changing etc.

In 2021, the Weilai ET7 was officially launched. As China's first model to be equipped with a lidar, it can achieve high-speed autonomous navigation assistance, and is also equipped with multiple high-precision cameras, millimeter wave radar, ultrasonic radar, and high-precision maps have started to be installed.

In the same year, Xiaopeng, and Huawei's collaboration with Ge Fox also officially announced or launched their own models equipped with laser radar, Chinese car companies' smart cars began to show their fangs.

Not just hardware, the route of Zhi Jia is actually full of variables. First of all, the uncertainty of high-precision maps. As the automatic navigation assistance function continues to iterate, different car companies are rapidly catching up in terms of performance on highways NOA, and most car companies' strategies for highway NOA have been achieved through perception hardware, body positioning system, combined with high-precision maps to achieve "automatic" driving.

Compared to high-speed scenarios, the biggest problem with urban scenes is that road conditions are much more complex, especially in terms of maintenance and construction, as well as information security around important units, which puts high-precision maps in an awkward position. The first awkward point is that the update frequency cannot keep up, frequent road repairs bring huge costs to the collection of high-precision maps, this cost pressure will eventually be passed on to car manufacturers, who also bear a huge burden.

Not only is the lidar technology troubled by cost and technological routes, but so too is the laser radar. The lidar route that was just launched two years ago has started to appear voices of abandoning laser radar again in recent year. However, compared with giving up high-precision maps, laser radar is troubled by decreasing costs and its own product characteristics, whether or not to use laser radar as a necessary hardware for intelligent driving, each car company needs to make a rational choice.

Epilogue: (Note: This is just a placeholder translation for an epilogue section in a book or other written work. The actual content of this section would depend on the specific source text.)

For ordinary users, we just hope to get a good and affordable solution. Looking back at the development end, hardware has also been divided into the laser radar camp that advocates "hardware won't break" and the visual dominant camp that advocates "watering". Of course, in a long period of time, the balance between cost and effect still can not be stable, we will talk about the most popular laser radar in the next issue.

EVNows

EVNows provides the latest news and updates on Chinese electric vehicles. Stay informed with comprehensive coverage of the EV market in China, including trends, insights, and innovations. Discover the future of electric cars with our up-to-date reports and analyses.

Popular Categories

Copyrights © evnows.com 2024. All Rights Reserved.