Level 2 partial assisted driving, including adaptive cruise control and lane centering assistance, has become standard configuration for most mid-range new cars, functioning as fatigue-reduction tools requiring constant driver supervision. Level 3 marks a fundamental liability shift: when the automated system is active within approved scenarios, manufacturers bear legal responsibility for accidents caused by system errors, rather than individual drivers. Germany, Japan and several U.S. states have finalized complete Level 3 regulatory frameworks, permitting Mercedes-Benz, Honda and other brands to launch road-legal Level 3 vehicles for highway use. These vehicles process data from LiDAR, millimeter-wave radar, surround-view cameras and high-precision maps to handle overtaking, lane changes, congestion following and emergency obstacle avoidance independently. Urban Level 4 fully autonomous robotaxis operate in geofenced downtown districts in cities such as San Francisco, Shanghai and Singapore, serving paid ride-hailing services without in-car human safety drivers. These fleets accumulate massive real-world driving data continuously to refine algorithm decision-making logic, yet unpredictable complex urban scenarios—jaywalking pedestrians, illegal parked vehicles, sudden construction zones and erratic cyclist behavior—still trigger frequent system disengagement, limiting expansion beyond restricted service areas. Safety remains the most heated public controversy surrounding autonomous vehicles. High-profile crash incidents in previous years raised public skepticism about sensor failure, algorithm misjudgment and adverse weather adaptability; heavy rain, dense fog and snow cover degrade LiDAR and camera sensing accuracy significantly. Ethical dilemmas also remain unresolved: when unavoidable collisions are imminent, how should autonomous systems weigh harm to passengers versus pedestrians, lacking universal global ethical legislation guidelines. Infrastructure incompatibility creates another major bottleneck. Most public roads were designed for human drivers, lacking standardized intelligent road sensors, real-time vehicle-to-everything (V2X) communication modules and unified high-definition map updating mechanisms. Upgrading nationwide road infrastructure requires astronomical long-term government investment. Cyber security threats also persist: malicious remote hacking of autonomous vehicle control systems could trigger large-scale traffic accidents, demanding robust end-to-end encryption and intrusion prevention architectures. Automakers adjust long-term strategies accordingly, phasing gradual stepwise autonomy rollout instead of rushing full self-driving launches. Short-term focus remains perfecting Level 3 highway commercialization and expanding geofenced Level 4 fleet operation scope, while policymakers update traffic laws, liability rules and data privacy regulations to match technological progress. Autonomous vehicles will reshape urban transportation eventually, but full nationwide mass adoption will likely take 10 to 15 more years of iterative improvement and systemic preparation.
