Is your self-driving car racist?!
For many, that question seems preposterous, but it is a valid one. That’s because the way in which obstacle avoidance works still has some flaws.
These flaws adversely impact some groups, especially those with darker skin. So, the truth is your self-driving car might be racist due to the AI technology it uses to “see”. Of course, Tesla and others are working hard to fix those issues ASAP.
Do you want to learn more about how and why this is the case? If so, then you first need to understand the tech that drives your car. Read on to get informed.
How Does a Driverless Car “See”?
Table of Contents
So, how does a driverless car “see”? When a human is behind the wheel, their eyes perceive the environment. You see the road, other cars, pedestrians, and more.
Your car “sees” the same things through the use of radar, lidar, and cameras. In this way, the car builds a map and places the car in it.
From there, it plans out a safe path based on roadways, changing lanes, and passing cars.
Map building and path planning are only half the battle, though.
Your car also needs to “see” objects on the road. Then, the car needs to know what to do to avoid that object. The car learns what to do by identifying the object and predicting what it will do based on AI learning.
Understanding the Flaws of Full Automation
Obstacle avoidance is by far the most difficult part for cars. We have to train the car’s brain to predict things accurately. This is, by far, where many of the flaws of full automation take center stage.
Useful information regarding image segmentation explains how predictions happen. First, countless data points get shown to the AI “brain”. The computer takes all that data, analyzes it, and uses it to predict future behavior.
It’s an eerie behavior that matches and exceeds human learning processes.
But, it’s not perfect.
If the right data isn’t put in the machine, then the brain won’t make the right choice. It still has trouble figuring out how to understand emergency vehicles, for instance. It also struggles to distinguish darker objects from lighter ones.
Self-Driving Car Technology is Improving Fast
The good news is that these flaws can get overcome.
All we need to do is input the right types of data into driverless tech AI systems. Over time, better and better decisions will be the outcome.
Obstacle Avoidance: What to Expect in 2022 and Beyond
Your self-driving car is a technological wonder. It’s powered by hundreds of years of evolution. When cameras “see” an object, it uses predicted behavior to guess how that object will act. In that way, the car’s AI ‘brain’ is always learning.
While that might sound scary, the car’s AI brain is less prone to errors than a human one!
Knowing what you know now, do you still trust self-driving cars? Will you be buying one anytime soon?
If you’re interested in reading more technological news, then we’ve got all the latest. Head over to our main tech page now to see more of our best articles.