⏱ 10 Min Read Traffic Safety & Autonomy File of the Week

The source of the problem was not driver inattention, but rather that the cameras and sensors—the AI-powered eyes of the vehicles—were essentially rendered "blind" under the torrential downpour. Lane lines disappearing under puddles, the water spray kicked up by the vehicle ahead, and heavy raindrops deflecting LiDAR lasers caused multi-billion-dollar AI algorithms to crash in seconds. So, who pays the bill for accidents occurring during this moment of technological blindness?

1. The Technological Dead End: Why Can't Machines See in the Rain?

For a modern autonomous vehicle to perceive the world, it relies on three primary sensory organs: Cameras (Vision), Radar (measuring distance with radio waves), and LiDAR (3D mapping with laser beams). On a sunny spring day, this trio works in a flawless symphony. However, when the August monsoons begin, nature is ruthless against the machine.

  • Cameras: Tasked with reading lanes and traffic signs. However, puddles on the asphalt create a mirror effect, making streetlights look like lanes (false positive). When raindrops hit the camera lens, the algorithm sees only a blurry smear instead of the vehicle ahead.
  • LiDAR (Laser): Creates a millimeter-accurate 3D map of the environment. But when there are dense water droplets in the air, the laser beams hit these droplets and bounce back. The AI, thinking there are massive, transparent, and shapeless obstacles in front of it (ghost objects), might suddenly execute a "Phantom Braking" maneuver in the middle of the highway.
  • Radar: The system least affected by rain; however, radar only knows that metal or an obstacle is there—it cannot distinguish what that obstacle is (a pedestrian vs. a flying plastic bag). Without confirmation from the cameras, radar alone cannot steer the vehicle.

The Critical Threshold: The "Eyes on the Road, Hands Free" Paradox

The definition of Level 3 autonomy is this: "Under appropriate conditions, the vehicle drives itself; the driver can take their hands off the steering wheel. However, when the system detects a problem, it issues a warning, and the driver must take over control within seconds." The crisis occurs exactly at this moment of "handover." When the system suddenly goes blind in heavy rain, it shouts, "Take control!" But is 2 seconds enough time for a human, who is traveling at 120 km/h and looking at their phone or multimedia screen, to regain control of a vehicle on a flooded highway? Accident reports prove that this time is insufficient.

2. Panic in the Insurance Sector: Who is to Blame?

Following the chain-reaction crashes in Seoul, the South Korean insurance sector and traffic courts found themselves facing an unprecedented mountain of files. Drivers and Original Equipment Manufacturers (OEMs) are blaming each other for millions of won in damage claims.

The Drivers' Defense: "The automaker sold me this vehicle as 'Autonomous'. If the system isn't going to work in the rain, why didn't it warn me, or why didn't it safely pull over and stop? By suddenly disengaging, it left me out of control in the middle of a puddle."

The Automakers' Defense: "Our user manual (EULA) clearly states that Level 3 autonomy should not be used in conditions of 'heavy rain, snow, and fog'. Furthermore, the system issued an audible warning 3 seconds before handing over control. The driver is at fault for not paying attention to the road."

"If an autonomous system only works on sunny, perfect days, it's not artificial intelligence; it's a luxury cruise control. If machines are going to give up in bad weather and throw the steering wheel to a terrified human, the promised safety revolution of this technology is a massive illusion."

Autonomy in Bad Weather and the Distribution of Legal Liability

Level 2 (Advanced Assistant)
Hands must be on the wheel. If an accident occurs in bad weather, All Blame is on the Driver.
Level 3 (Center of the Crisis)
Hands free. If the system warns and disengages in bad weather, the Gray Area (Insurance Lawsuits) begins.
Level 4/5 (Full Autonomy)
The vehicle drives completely on its own. If it crashes in bad weather, All Blame is on the Manufacturer. (Not yet legal in South Korea)

3. Conclusion: New Regulations and the Future

This monsoon crisis in August pushed the South Korean Ministry of Land, Infrastructure and Transport to prepare an emergency action plan. According to the new draft, when the AI systems in cars detect "heavy rain" based on the speed of the wipers and external ambient moisture sensors, they will be forced to automatically lock out (prevent the engagement of) Level 3 autonomous driving.

The future of autonomous vehicles lies not only in making algorithms smarter but also in ensuring that machines "know when they have reached the limits of their capacity." If AI, which sees the world better than a human on a sunny day, turns into a child seeing only blurry pixels in a monsoon rain; the legal system must issue the bill not to the human who over-trusted the technology, but to the tech giants who sold that trust and shirked responsibility.