Legal Framework for Self-Driving Cars: Navigating the Theoretical and Russian Frontier

TL;DR

As self-driving technology races toward SAE Level 5 (full autonomy), the legal systems that govern our roads are struggling to keep pace. This paper explores the "Legal Framework for Self-Driving Cars," focusing on the Russian case study. It argues that the biggest hurdle isn't the AI's ability to drive, but our legal inability to define the AI as a "driver" and assign liability when things go wrong.

The Core Friction: Human Law vs. Machine Logic

The fundamental problem is that our entire global traffic architecture—anchored by the Vienna Convention on Road Traffic—is built on the assumption of human agency. Article 8 explicitly states that "every moving vehicle shall have a driver."

In a world of Tesla Autopilots (Level 2) or Waymo taxis (Level 4/5), this definition cracks. If there is no steering wheel or pedals, can a human even be called a "driver"? The authors highlight that the current legal status in Russia and much of the world treats AVs as "experiments" rather than established practice.

Methodology: Mapping Autonomy to the Law

The authors use the SAE J3016 classification as a bridge between technology and law. They argue that while Levels 1-3 are manageable by tweaking "driver assistance" rules, Levels 4 and 5 require a paradigm shift.

Figure 1: The paper discusses the transition from human-controlled (SAE0) to fully autonomous (SAE5) systems and the legal vacuum it creates.

The Case of Russia: From Policy to Public Roads

Russia has transitioned from mere "mentions" of AVs in 2010 to active road testing. Key milestones include:

• Government Decree No. 1415 (2018): Launched experiments in Moscow and Tatarstan. It introduced terms like "automated driving system" into Russian law for the first time.
• The "AutoNet" Roadmap: A strategic plan aiming for full integration by 2035, focusing on cybersecurity, Big Data, and digital mapping.
• Infrastructure Gaps: The authors note that Russian roads face unique challenges—poor markings and extreme weather—that require more complex AI and, consequently, more robust technical regulations.

The Liability Paradox: Who Pays?

The paper delves into three potential liability models for a world without human drivers:

1. Producer Responsibility: Treating a crash like a "malfunctioning toaster." However, software updates (SOTA) make it hard to determine when the "defect" was introduced.
2. Owner Liability: The owner of the "high-risk tool" pays, regardless of who was driving.
3. Infrastructure Liability: If a "smart road" sends a wrong signal to a car, the municipality might be at fault.

Figure 2: Analysis of the regulatory trajectory in Russia compared to US/EU standards.

Critical Insight: The "Robot Driver" Status

A fascinating takeaway is the contrast in international approaches. While Europe (and Russia) focuses on adapting the human driver concept, the US NHTSA has already signalled a willingness to recognize Google's AI software itself as the driver. This "Equalization of Rights" between human and software may be the shortcut needed to solve the liability puzzle.

Conclusion & Future Outlook

The paper concludes that while Russia is making strides, the legal "readiness" is still low. Future work must bridge Artificial Intelligence ethics with Insurance law. We don't just need cars that can see; we need laws that can "see" the AI as a responsible entity.

Takeaway: The commercialization of AVs will not be determined by Lidar costs, but by the speed at which we can rewrite the 1968 Vienna Convention.