​This Is How Bad Self-Driving Cars Suck In The Rain

We may earn a commission from links on this page.

Hyundai recently held its Future Automobile Technology Competition in South Korea, pitting 12 teams against each other to see which autonomous tech comes out on top. Only four teams made it to the final round, and on the second day, disaster. A strange liquid fell from the sky.

This and other environmental anomalies – snow, fog, smoke, sand, locusts – continue to be an issue for self-driving cars, but it's never been so clearly illustrated than with the two clips below.

Here's the first day, with a dry course and the sun out:

Then the second day, just after it rained, same car, same equipment, same software:

For the full effect, take a tip from IEEE and click here for the YouTube Doubler version to see them side-by-side.

Advertisement

The issues with the KAIST Unmanned Systems Research Group's car were numerous, but the biggest problems had less to do with the slippery road surface and more to do with the visual systems. Those cameras and LIDAR arrays are dependent on a clear view, and with the angle of the car shifting and the direction of the sun, the sensors fail to pick up everything from street signs to lane markings and even pedestrians. And it just keeps getting worse.

Advertisement

The team has to hit the emergency stop button at least twice, veers onto the side of the road, doesn't see a curb and almost slams into a light pole, and then smacks into a barrier when parking.

Advertisement

But even during the dry run the car hits the barricades, so even in perfect conditions there are kinks to work out.

And it could've been worse:

Hey, at least it signaled.