How do these self-driving cars crash? Well, in the case of the incidents above, it’s human error — and not at all the error of the self-driving algorithms that direct the car. Instead, these accidents are the result of people rear-ending these cars that have been either stationary or traveling at five miles an hour or fewer.

Read More: Self-Driving Taxis: the Cabs of the Future

Redesigns by Google and designs from other self-driving car companies have disguised these sensors to better help the cars blend in with general traffic and maybe help avoid more rear-end collisions from distracted drivers in the future.

With laser, radar, sonar, and video sensors on these cars, there’s a 360-degree view of the road that’s taken into consideration. With more than a million and a half logged miles for these self driving cars, there haven’t been any serious accidents. In fact, it is believed that the cars have even helped to prevent some accidents from happening in the first place.

For now, humans are seated inside, ready to take over when necessary. Sometimes, it is necessary. There may be problems with the sensors or something going wrong with the steering or braking, in which a human driver needs to quickly take control.

Analyzing Scenarios
In the case of the self-driving Google cars, the driver is given audio and visual signals that they need to take over. No matter what conditions are on the road, or why a human takes over the wheel, these scenarios are played out in Google on a powerful software.

This software is able to predict the reaction of others in regard to the test scene. It can then figure out whether or not the car was taken over because it was going to cause an accident or if the human driver took over for a reason other than one pertaining to safety.

Self-driving cars could open up a world of possibility. Elderly or handicapped drivers, who can no longer drive on their own, could one day still maintain a sense of independence, getting from here to there in a self-driving car.