L3 Autonomous Driving Approval: Automakers' Tech Route and Regulatory Balance Strategies

Let me provide you with an in-depth analysis of the key strategic issue of how automakers balance technical routes and regulatory requirements after the implementation of L3 autonomous driving access approval.

represent the thinking of the tech radical camp:

• Technical Feasibility
  : L3 systems require human-machine co-driving handover, which has the problem of ambiguous responsibility boundaries, while L4 can completely avoid this issue through scenario limitation
• Cost Efficiency
  : Dual-line parallel investment is huge; concentrating resources to break through L4 has more strategic value
• Market Positioning
  : L4 has clear commercialization scenarios (Robotaxi, unmanned delivery, etc.), while L3 faces challenges in user expectation management

: From the rollout of FSD V14.2 in North America and Austin Robotaxi tests [1], the pure vision solution + end-to-end neural network is accelerating towards L4 evolution. Tesla was even required by the California DMV to stop using related promotions due to its marketing of the “autonomous driving” term [2], reflecting regulatory vigilance against “full autonomous driving.”

are based on regulatory reality:

• Legal Watershed
  : L3 is a key node for responsibility division—drivers are responsible for L3 and below, while automakers are responsible for L4 and above. The Ministry of Industry and Information Technology (MIIT) issuing L3 access approval and the special license plate “Beijing AA0001Z” marks the official opening of this responsibility watershed by the regulatory system
• Data Accumulation
  : L3 provides large-scale real road data, laying the foundation for L4 algorithm training
• Technical Ladder
  : The gradual path from L2→L2+→L2.9→L3 conforms to the law of engineering iteration

China’s L3 access approval implementation reflects the regulatory philosophy of

:

• Product Access
  : MIIT conducts L3 function certification for vehicle models (e.g., Changan Deepal SL03, BAIC Arcfox Alpha S)
• License Plate Management
  : The special license plate “Beijing AA0001Z” facilitates identification and tracking by regulatory authorities
• Liability Insurance
  : Corresponding insurance products need to be matched to cover liability risks in L3 scenarios

• U.S. California Model
  : Requires Tesla to stop using the “self-driving” term [2], showing strict control over marketing promotions
• European Model
  : Emphasizes the strict technical requirements of UN-R157 regulations for ALKS (Automated Lane Keeping System)
• China Model
  focuses more on a trinity management system of “access approval + special license plate + liability insurance”

Technology Roadmap:
Passenger Car Market: L2+ → L2.9 (Conditional L3) → Scenario-Limited L4
Commercial Operation: L4 Robotaxi/Unmanned Delivery (Area-Limited)

: Leading enterprises with strong R&D capabilities

• Xpeng
  : While advancing L4 Robotaxi, it can first meet user needs through L2+ products to avoid regulatory risks
• Tesla
  : Advance L4 Robotaxi in North America; in China, it can adapt to L3 access requirements as a transition

: Hardware pre-installs L4 capabilities; software is gradually unlocked via OTA

• Current
  : Launch under the name of L2.9 (e.g., Tesla FSD, Xpeng XNGP)
• Regulatory Compliance
  : Apply for L3 access approval, but clarify user responsibility boundaries
• Future Evolution
  : After regulations mature, upgrade to real L4 via OTA

: Avoid responsibility traps while reserving technical space for the future

China Market: Strictly adhere to L3 access framework, accumulate data, improve regulatory system
Europe & US Markets: Directly advance L4 commercialization (e.g., Waymo model)

: Multinational enterprises with global layout (e.g., Tesla, Xpeng Internationalization)

The core contradiction of L3 is the “responsibility vacuum period”:

• Takeover Scenario
  : L3 requires drivers to take over when requested by the system, but how to define collision responsibility before takeover?
• Solution
  : Automakers need to cooperate with insurance companies to develop “L3 exclusive insurance”, learn from Tesla Robotaxi’s business model of charging $150 cleaning fee [3], and establish supporting service systems

L3 operation generates massive driving data:

• Data Localization
  : Must comply with China’s “Data Security Law” requirements; data stored domestically
• Privacy Protection
  : In-car sensors such as cameras and microphones must comply with personal information protection norms

• V2X Construction
  : L3/L4 require vehicle-road collaboration; China is accelerating the advancement of intelligent road infrastructure
• Geofencing
  : Initial L3/L4 may be limited to specific cities or roads (e.g., Beijing Yizhuang, Shanghai Jiading)

1. Short-Term Strategy
   : Accumulate users and data through L2.9 products to avoid L3 responsibility traps
2. Mid-Term Layout
   : Advance L4 commercialization in limited scenarios (e.g., parking lots, highways)
3. Long-Term Goal
   : After the regulatory framework matures, fully switch to L4

1. Leverage First-Mover Advantage
   : As the first batch of automakers to obtain L3 access, establish technical and data barriers
2. Improve Supporting Services
   : Cooperate with insurance companies and government departments to build an L3 ecosystem
3. Technical Reserve
   : Develop L4 technology in parallel to prepare for future switching

1. Dynamically Adjust Access Standards
   : Update L3/L4 certification requirements in a timely manner according to technological evolution
2. Pilot Sandbox
   : Establish L4 test zones in specific areas (e.g., Beijing, Shanghai)
3. Liability Insurance Innovation
   : Promote the insurance industry to develop exclusive products for autonomous driving

:

• L3 models will be gradually mass-produced, mainly limited to highway scenarios
• L4 Robotaxi will be piloted in limited areas (e.g., Tesla Austin, Waymo Phoenix)
• Regulatory framework will be gradually improved, and responsibility boundaries will be clarified

:

• Technology matures and costs decrease; L4 begins large-scale commercialization in specific scenarios
• L3 gradually evolves to L4, and responsibility is fully transferred to automakers
• Industrial ecosystem is reconstructed; cars transform from transportation tools to mobile service terminals

Automakers should not simply choose to “skip L3” or “stick to L3”; instead, they should adopt the

:

• Technical Aspect
  : Hardware pre-installs L4 capabilities; software is gradually unlocked
• Market Aspect
  : Passenger cars take the L2.9→L3 route; commercial operations take the L4 route
• Regulatory Aspect
  : Actively cooperate with the L3 access framework, participate in standard setting, and pave the way for L4

: The real value of L3 is not in the technology itself, but in that it is a . Automakers need to break out of the pure technical perspective and plan autonomous driving strategies from the height of industrial ecosystem reconstruction.

[1] Tesla Ramps Up Robotaxi Push Amid 2025 Deadline and Delays - WebProNews (https://www.webpronews.com/tesla-ramps-up-robotaxi-push-amid-2025-deadline-and-delays/)
[2] California Demands Tesla Stop Calling Its Cars ‘Self-Driving’ - SlashGear (https://www.slashgear.com/2059436/tesla-california-full-self-driving-false-marketing/)
[3] Tesla Will Charge You $150 If You Puke In A Robotaxi - InsideEVs (https://insideevs.com/news/782837/tesla-cybercab-150-cleaning-fee/)