From knock-off old-man's electric vehicles to industrial giant

Chapter 183 Bottlenecks of Domestic LiDAR
In fact, as soon as he landed today, Xu Yi accompanied him to inspect the actual Xingchen S7 vehicle and production line.

We then discussed our views on autonomous driving and information about the launch of the next generation of new cars.

Therefore, Wu Xinzhou was very tempted by the "cake" that Xu Yi painted!
If Xingchen's next car can break into the price range of over 300,000 yuan, even if it only achieves annual sales of 80,000 units.

It will surely become a benchmark for verifying high-end intelligent driving in China.

"Welcome to Star Technology! You won't regret this decision!"

Xu Yi immediately revealed an even brighter smile.

The two stood up and shook hands formally.

As originally planned, the position of Vice President of the Autonomous Driving Department and head of the team has been confirmed.

The last weakness of Xingchen Motors has been addressed.

"In terms of intelligent driving solutions, we plan to choose the NVIDIA Drive PX 2 platform as the main platform. The goal of this next-generation vehicle is to enable navigation functions on all highway sections, achieving intelligent driving freedom in highway scenarios."

As soon as he sat down, Xu Yi couldn't wait to speak.

Currently, there are only a handful of solutions available worldwide for the intelligent driving industry.

Besides Israel's "Mobileye solution", the only options left are Nvidia's solution and Texas Instruments' solution.

Texas Instruments' TDA2x chip has a computing power of only 3 TOPS and can only support basic L2 functions.

According to industry standards, 1tops means that one trillion operations can be performed per second.

It looks like a lot?
In fact, to achieve true intelligent driving capabilities, chip computing power is definitely one of the core factors, and in the later stages, it will be increased to 1000 TOPS.

of course.

Current industry level.

A computing power of a few dozen TOPS is already top-notch!
“Mr. Xu, I didn’t expect your ideas to coincide with mine. Although Mobileye’s supporting products are good, overall, NVIDIA Drive PX 2 has a higher ceiling and is more suitable for us to develop our own algorithms.”

Wu Xinzhou's eyes lit up.

In the current industry environment, Mobileye can be considered a leader in Level 2 computing.

Although Nvidia's chip has three times the computing power of EyeQ4, it cannot provide deep collaboration with the corresponding perception algorithms.

"Hey—I just don't want to do business with Israeli companies."

Xu Yi waved his hand.

As for Mobileye, well, those who know, know.

Don't be fooled by our current status as a top L2 autonomous driving supplier; history has shown how fast we can catch up.

These Israeli suppliers are so difficult to deal with that you can't bribe them with money; you have to beg them to release their software and hardware. It's torture for both the purchasing side and the R&D engineers. Anyone who has ever dealt with Mobileye has learned to avoid them.

Tesla's first-generation HW1.0 Autopilot used Mobileye, but last year HW2.0 switched directly to Nvidia's chip.

However, domestic companies, in pursuit of quick entry and implementation of solutions,

The first-generation NIO and the first-generation XPeng both chose the EyeQ4 + ZF S-Cam 4.0.

Unfortunately, Mobileye has black-box limitations, making it impossible for it to fully open its underlying interfaces.

This meant that for the next two or three years, they were inherently no match for Tesla.

Xingchen uses the NVIDIA platform, which puts them on the same starting line as Tesla's hardware.

……

With Wu Xinzhou joining the team.

Starry Sky's own "autonomous driving vehicle" can now officially enter the launch phase.

the next day.

Xu Yi then took advantage of the fact that the other party had not yet returned to China to take Wu Xinzhou to inspect the lidar supplier.

At present, in terms of hardware for intelligent driving, in addition to chips, cameras, millimeter-wave radar, etc., LiDAR is an essential hardware configuration for running the early multimodal algorithm data of autonomous vehicles.

The concepts and technology verification of driverless cars and autonomous driving had already begun abroad as early as 04.

Take, for example, the once-popular "DARPA Challenge".

While domestic gasoline-powered cars were still struggling to be manufactured, other countries had already begun challenging driverless cars to travel 142 miles in the desert. Due to technological bottlenecks, only one car ultimately completed 11 miles.

However, in the second year...

A team from Stanford University won the championship with a time of 6 hours and 54 minutes, proving for the first time that lidar can effectively perceive terrain.

After that, the LiDAR approach became unstoppable.

Almost all driverless cars and challenges have become essential hardware.

After all, the 3D sensing capability of cameras is too weak, and the resolution of millimeter-wave radar is low.

Only point cloud data from lidar can accurately perceive the "position, distance, and shape" of obstacles.

Most of the domestic startups capable of manufacturing lidar were founded in 13-14.

But the efficiency of catching up is not slow.

For example, their target this time is Hesai Technology.

This year, a new 40-line LiDAR was launched, which is very close to the 64-line LiDAR of the leading foreign companies.

Of course, it is still a mechanical radar in form, and the cost is astonishingly high, reaching 10 yuan per unit. That's enough to buy a car, and only self-driving cars used for training algorithms can afford it. Ordinary consumers simply won't buy it.

However, this is much better than the price of three times that of imported goods.

Hesai Technology is headquartered in Shanghai.

The group arrived by high-speed rail and then met with the senior management team of Hesai Technology.

"Mr. Xu, hello..."

Hesai Technology CEO Li Yifan, along with the other two founders, are all born in the 85s, making them younger than most of the supply chain executives Xu Yi has met.

of course.

This is also a purely research and technology-driven company, and its three founders are all geniuses in the fields of motion control and sensors.

When a "new customer" arrives, Li Yifan and the entire company team naturally welcome it.

But there are also clear new anxieties.

really.

Xu Yi's next sentence hit the nail on the head regarding the core of this anxiety.

"Mr. Li, we've already discussed this beforehand. Xingchen Automotive plans to order a batch of 'Pandar40' units for training multimodal algorithms for our first batch of driverless cars. Price is not an issue, and we hope you can support us."

Pandar40, also known as Hesai's latest 40-line mechanical lidar, is priced at a full two-thirds lower than its overseas counterparts.

Naturally, this became Xingchen's first choice for procurement.

Li Yifan pondered, "I wonder how many Starry Sky plans to take?"

Xu Yi: "At least 200, for the first batch of unmanned test vehicles, we have a huge demand."

His tone was crisp and decisive.

He and Wu Xinzhou had discussed it on the way here.

To collect data on a large scale and in different city sections, and to "overtake on curves," a few or even a dozen vehicles are clearly not enough.

There should be at least seventy or eighty cars, or even a hundred cars.

Considering the losses of mechanical lidar and the scenario where a single vehicle carries two lidars to collect data, at least 150 to 200 lidars would be needed to support this test.

As soon as these words came out.

Li Yifan and the other two founders immediately gave wry smiles.

"Mr. Xu, we are afraid that we will have difficulty meeting the current needs of your Starry Sky."

Although the Pandar40 has entered mass production, the total annual output is less than 400 units. A large portion of our orders are already booked, and we're afraid we can only squeeze out a few dozen units this year at most. Production capacity is extremely limited, and we want to supply as much as possible.

Li Yifan's words were very tactful, clearly indicating that he did not want to offend Xingchen Automotive, a potential "major customer" in the future.

……

(End of this chapter)