Future technology with no future? Is Lidar retreating in China?

Related Vendor

Wuxi InfiMotion Technology Co., Ltd.

For years, Tesla founder Musk has criticized Lidar, while this technology has been largely accepted by the rest of the automotive world. However, a turnaround is now at least noticeable in China. Is Lidar too unattractive for a real breakthrough?

Did Elon Musk turn out to be right? The Tesla founder has long been an opponent of Lidar. His judgments on the radar devices range from "unnecessary" for autonomous driving to "futile effort" to a "crutch". However, the industry, especially in China, took a different path.

Recently, however, some new car models are coming onto the market in the People's Republic that do not have Lidar. The AITO M7 Pro by Seres and Huawei, the Mona M03 by Xpeng, and the "DeepAL S7 Qiankun" by Changan Automobile, similar to Tesla, rely on "Pure Vision," which means autonomous driving functions using cameras and algorithms.

While most Chinese car manufacturers currently still rely on Lidar, these new models signal a first rethinking in the industry. Advances in hardware and particularly in data processing using artificial intelligence have unexpectedly rapidly improved the performance of Pure Vision systems, experts say.

Additionally, the omission of Lidar allows for a reduction in costs. The new electric car from Xpeng, Mona M03 Max, which was launched in China in August, is a vivid example of this. Xpeng markets the model with the claim it is the cheapest "high-end car with smart driving features," including ADAS, that has ever existed. For a price of less than 20,000 euros, buyers in China get an electric car that is equipped with a navigation system, advanced driving assistance, and other comforts like parking aids in many cities and on highways.

Lidar? We don't need it.

The Xpeng Mona M03 Max does not have a Lidar device. The car is equipped with twelve cameras, twelve ultrasonic radar devices, and three millimeter-wave radar devices. The advanced ADAS functions are made possible thanks to Xpeng's new "AI Eagle Eye Vision Solution," which uses AI to enhance the processing of visually acquired information from front and rear cameras.

The result, similar to Tesla, is an "End-to-End" solution (E2E) for autonomous driving functions and ADAS. E2E means that the information gathered by cameras and other sensors is processed using deep learning within the car itself into the output required for navigating, braking, and accelerating.

In the Chinese industry, it has been closely observed that Mobileye, the leading Israeli provider of autonomous driving solutions, which was bought by Intel a few years ago, has also just made a U-turn regarding Lidar. On September 9, 2024, Mobileye announced that it will forego the development and production of FMCW-Lidar. About 100 jobs in the development department involved will be cut.

This decision by Mobileye "partially vindicates Elon Musk," comments the specialist portal Teslerati.com. And in the Chinese trade press, the old debate has reignited over which of the two technological routes for autonomous driving, E2E or "Pure Vision" on one hand, and Lidar-supported systems on the other, will prevail from now on.

"Will the E2E route of visual smart cars kill Lidar?" asks the Chinese car newspaper Zhongguo Qiche Bao. A number of Chinese OEMs and ADAS system providers, including Baidu, Huawei, Li Auto, Nio, and SenseTime, have already "begun to quickly switch to the E2E technological route," writes the car newspaper.

Parallel development

Many analysts in China currently assume that both solutions—E2E and Lidar—will continue to be developed in parallel, sometimes even by the same car manufacturer, depending on the price and positioning of the respective car model. While prices for Lidar devices in China have recently fallen to less than 200 US dollars per unit, the devices will probably continue to be installed only in the mid and premium class, while being omitted in affordable cars as before.

The reasons for this are: On one hand, Lidar devices are good for giving car buyers a sense of security. They allow for additional validation of data obtained from cameras. On the other hand, the requirements for semiconductors—and the associated costs—increase when a manufacturer relies entirely on "Pure Vision." (sb)

*The author is Managing Director at AsiaWaypoint, Beijing 

Shift away from China

After 25 years: IBM closes two R&D centers in China

Legacy chips are top, modern chips a flop?

Car chips are the Achilles heel of Chinese e-mobility