National People's Congress Representative, Xiaopeng Motors Chairman and CEO He Xiaopeng: Accelerate the transition of autonomous driving regulations and technology from L2 to L4

Financial Times Reporter Wang Xiaowei

At the 2026 National Two Sessions, NPC Deputy and XPeng Motors Chairman and CEO He Xiaopeng proposed three suggestions focused on autonomous driving, embodied intelligence, and low-altitude economy. He believes 2026 will be a year of continuous technological breakthroughs and a critical point where institutional supply and industrial scale need to be aligned. He hopes that scientific and technological achievements can truly move out of laboratories and become accessible realities in city skies, on roads, and in home scenarios.

Regarding accelerating the transition of autonomous driving technology from L2 to L4 and improving regulations and management policies, He Xiaopeng suggests that promoting policy and regulatory systems to leap from L2 to L4, facilitating rapid technological iteration and large-scale commercial use, will help China convert its advantages in L2 into a competitive edge in the L4 autonomous driving era.

He Xiaopeng offers four recommendations: First, maintain the stable operation of the L2 safety supervision system while promoting the policy shift from L2 to L4, simplifying the intermediate L3 stage. Second, gradually clarify the registration and passage management system for L4 autonomous vehicles, and promote compliant deployment of L4 vehicles nationwide. Third, conduct traffic regulation adaptability assessments, optimize traffic behavior norms suitable for both human and machine driving under safety conditions, considering autonomous driving features. Fourth, grant local pilot management rights for L4 unmanned driving applications in specific scenarios, allowing some cities with mature basic conditions to conduct pilot projects in low-risk scenarios, gradually forming replicable and promotable experiences.

He Xiaopeng believes that advancing autonomous driving from L2 directly to L4, bypassing L3, and accelerating breakthroughs in key institutional bottlenecks that restrict industry development will give China a strategic advantage in the global competition of intelligent connected vehicles and autonomous driving industries.

In embodied intelligence, He Xiaopeng advocates encouraging accelerated research and commercialization of high-level intelligent humanoid robots with “edge-side local brains.” He analyzes that most current humanoid robots in China are controlled by software rules, demonstrating strong capabilities at the motion control system level, but lacking industry advantages in the collaboration system between the “brain” (autonomous thinking and decision-making) and the “cerebellum” (motion control), as well as in scene task generalization and commercialization prospects. Compared to this, high-level intelligent humanoid robots driven by physical-world large models, capable of autonomous perception, decision-making, and execution with generalization ability, are more conducive to widespread application in industrial, commercial, and even home scenarios, offering broader commercial value.

He Xiaopeng suggests that developing architecture and training for local large models, data collection and synthesis, and related computing power investments require significant resources. He recommends targeted R&D incentive policies, such as establishing national-level special R&D funds and strengthening full-chain tax policies, to support and guide the development of high-level intelligent humanoid robots, promote local deployment of large models in humanoid robots, and accelerate technological breakthroughs and commercialization. Additionally, he advocates speeding up the construction of standards for humanoid robot intelligence, referencing automotive autonomous driving classification standards, establishing standards, technical norms, and supporting requirements for humanoid robot intelligence, and clarifying core indicators such as computing power, data, application scenarios, and training intensity to provide unified basis for technological R&D, industry deployment, and regulatory oversight.

He also recommends optimizing low-altitude airspace management and tax policies for the flying car industry. He suggests decentralizing low-altitude airspace management authority, establishing a collaborative management mechanism involving military, government, and civilian sectors; selecting regions with mature low-altitude economy industries, excellent airspace resources, and relatively complete infrastructure to pilot refined low-altitude airspace management, and appropriately delegating management authority to sub-provincial or higher-level local governments to fully motivate local government initiatives in developing the low-altitude economy.

Regarding tax policies for the flying car industry, he proposes clarifying the classification of compliant aircraft that meet airworthiness standards based on their core attributes and functions, implementing phased tax reduction and exemption policies to reduce R&D, manufacturing, and market promotion costs, and stimulate market demand. He also suggests establishing a dynamic adjustment mechanism to gradually optimize the intensity and duration of tax incentives, ensuring smooth policy support and market operation transition from policy-driven to market-driven development.

(Edited by: Wen Jing)