Baidu forecasts 2025 L2+ intelligent driving penetration

  • May 1, 2023
  • Steve Rogerson

Chinese internet giant Baidu foresees that by 2026, the market penetration rate of smart car models equipped with L2+ level intelligent driving, valet parking, highway navigation pilot and city navigation pilot will exceed 15%.

On the eve of last month’s Auto Shanghai, Baidu released products covering intelligent driving, cabin and maps. Additionally, Baidu introduced four capabilities for car manufacturers in a white paper.

Rob Chu speaking at Auto Shanghai.

At the conference, Rob Chu, Baidu corporate vice president, announced that Black Sesame Semiconductor would become the latest chip partner to Baidu Apollo Smart Driving. Additionally, Apollo Highway Driving Pro will be deployed on the Black Sesame’s Huashan II A1000 chip computing platform, which will be officially released in Q3 this year.

Baidu predicts that by 2026, the market penetration of models equipped with the L2+ advanced intelligent driving, valet parking highway navigation pilot and city navigation pilot will exceed 15%, ushering in a transition period in the intelligent vehicle industry.

“In 2026, when intelligent cars become more prevalent, new customers will be less likely to consider cars without intelligent driving capabilities,” said Zhenyu Li, senior corporate vice president of Baidu.

Chu added: “Whoever can take the lead in providing consumers with a safe and secure intelligent driving experience will likely have a head start in the second half of the intelligent car competition.”

Baidu Apollo launched its flagship intelligent driving offering, Apollo City Driving Max, which is equipped with dual Nvidia Orin X with 508Tops arithmetic power. This offers users in China a coherent experience on urban roads using pure vision perception. The current product definition also includes lidar to achieve true perception redundancy. The product uses a lightweight high HD map that is nearly 80% lighter than the traditional HD maps, enabling rapid urban generalisation at a more reasonable cost.

Baidu’s Apollo Highway Driving Pro integrated auto-navigation plus valet parking has also completed an algorithm platform upgrade. With the latest single TDA4-VH platform, the AI and CPU computing power usage is 50% less than the previous TDA4-VH platform. It is expected to enable more intelligent driving and parking functionality at lower computing power and lower cost.

Baidu Apollo and Voyah, an EV manufacturer backed by Dongfeng Motor, also announced the Voyah Free, an all-electric and plug-in-hybrid five-seater SUV, will be equipped with Apollo Highway Driving Pro, which includes assisted driving functions such as highway navigation pilot, city autopilot and efficient parking in all scenarios.

In addition, Baidu Apollo also upgraded its Apollo Parking autonomous parking, which has a success rate of over 99% in the top five most commonly faced parking scenarios and takes only 28s to complete perpendicular parking, reducing the difficulty of parking for novice drivers.

Baidu Apollo has upgraded its in-vehicle navigation maps with different levels of quality, performance and real-time data. The newly released shared autonomy maps provide a human-vehicle interface and enhance users’ trust in smart driving. Simultaneously, Baidu’s HD maps have been upgraded to OneMap, which can quickly update map data in real time.

Baidu has integrated agile development with the traditional V-model followed by ASpice and passed the organisation-level assessment certification of ASpice CL3, becoming the first tier-one provider to achieve a deep integration of the two and be certified by an authoritative body.

Baidu Apollo released the “Baidu Apollo Self Driving Openness” white paper, announcing four key capabilities: open experience definition, open independent experience evolution, open full-cycle OTA services, and open co-create with car manufacturing partners. Baidu’s goal is to help car manufacturing partners achieve a leading position and commercial success in intelligent driving, while ensuring quality end-user experiences.