The Artificial Intelligence Behind Self-Driving Cars
Computer vision refers to the ability of a computer to “see” and understand what is happening in a given environment. Deep learning refers to algorithms that can learn and make predictions about data through experience.
>Deep learning is part of a broader family of machine learning methods based on artificial neural networks with representation learning. Learning can be supervised, semi-supervised or unsupervised.
By combining the two technologies, self-driving cars could mimic human behavior by being able to see and process images in real time. The AI software would then analyze these images and make decisions based on what it saw; such as whether it was safe to change lanes or turn left at an intersection.
How Self-Driving Cars Work
Self-driving cars use GPS, sensors and computer vision to help guide the car. When you hit the gas pedal in a typical car, it takes your foot off the brake. That lets the car accelerate when it is safe to do so. But a self-driving car doesn’t take your foot off the brake; instead, it keeps tabs on everything with its cameras and sensors—and once you’re ready to go, it accelerates and moves forward.
Use GPS (Global Position System). GPS is a network of satellites that uses radio signals to tell each other where they are and what time it is at that moment. This helps computers pinpoint their location anywhere on Earth at any time. In a self-driving car, this system makes sure the car’s position can be accurately determined so it knows exactly where it should go next.
Google’s Self-Driving Car
Google’s Self-Driving Car: The company first began their research on self-driving cars in 2007, with the Stanford Racing Team, who won a 2006 DARPA Grand Challenge using their robotic vehicle Stanley.
The Google car uses $70,000 worth of equipment that is mounted on the roof; this consists of video cameras, radar sensors and a laser range finder. It also works by combining data from these sensors with high resolution maps that contain information about fixed objects like curbs and signs. A learning algorithm helps the car predict what other drivers are doing so as to avoid accidents.
Since 2011, Google has been testing three generations of prototype vehicles and they also have a fleet of six Lexus RX450h SUVs which were modified to drive themselves. By 2012, it was estimated that Google had spent more than 300 man years creating its driverless cars and in August 2015 alone, their fleet drove 70 000 miles autonomously without human intervention whatsoever!
How Far Are We From Fully-Autonomous Vehicles?
While you’ve probably heard about all the hype surrounding the autonomous car, there’s still a bit of confusion about how close we are to seeing these cars on the road.
The truth is that we’re not at full autonomy just yet. And although the technology is getting better every day, it will take quite a bit of time before they’re widely available.
Google has been testing its self-driving car in select cities such as Mountain View and Austin, but there are still limitations on where these cars can travel, as Google keeps them around 35 miles per hour and only drives in sunny weather. While this may seem like an arbitrary restriction, it highlights one of many challenges facing autonomous driving: unpredictable situations require human intervention.
There’s also no self-driving car currently on the market. In fact, some automakers contend that fully autonomous vehicles aren’t likely to be available for another 10+ years.
There are a whole lot of steps between where we are today and complete autonomous mobility for vehicles.
- 10 Steps to Complete Autonomous Mobility
Not all of the developments are yet in place, but there will be. The road to autonomous mobility will take more than an SUV with some cameras and sensors. In fact, it could take more than a decade or more before we reach autonomous vehicles on our roads. Technologists need to create networked cars, safe algorithms for self-driving cars and technologies to manage driverless vehicles. If that sounds abstract, don’t worry—we’re going to walk you through the full 10-step list below from creation of a working infrastructure (including data collection) onward through regulatory frameworks and legal implications.The Artificial Intelligence Behind Self-Driving Cars
Computer vision refers to the ability of a computer to “see” and understand what is happening in a given environment. Deep learning refers to algorithms that can learn and make predictions about data through experience.
>Deep learning is part of a broader family of machine learning methods based on artificial neural networks with representation learning. Learning can be supervised, semi-supervised or unsupervised.
By combining the two technologies, self-driving cars could mimic human behavior by being able to see and process images in real time. The AI software would then analyze these images and make decisions based on what it saw; such as whether it was safe to change lanes or turn left at an intersection.
How Self-Driving Cars Work
Self-driving cars use GPS, sensors and computer vision to help guide the car. When you hit the gas pedal in a typical car, it takes your foot off the brake. That lets the car accelerate when it is safe to do so. But a self-driving car doesn’t take your foot off the brake; instead, it keeps tabs on everything with its cameras and sensors—and once you’re ready to go, it accelerates and moves forward.
Use GPS (Global Position System). GPS is a network of satellites that uses radio signals to tell each other where they are and what time it is at that moment. This helps computers pinpoint their location anywhere on Earth at any time. In a self-driving car, this system makes sure the car’s position can be accurately determined so it knows exactly where it should go next.
Google’s Self-Driving Car
Google’s Self-Driving Car: The company first began their research on self-driving cars in 2007, with the Stanford Racing Team, who won a 2006 DARPA Grand Challenge using their robotic vehicle Stanley.
The Google car uses $70,000 worth of equipment that is mounted on the roof; this consists of video cameras, radar sensors and a laser range finder. It also works by combining data from these sensors with high resolution maps that contain information about fixed objects like curbs and signs. A learning algorithm helps the car predict what other drivers are doing so as to avoid accidents.
Since 2011, Google has been testing three generations of prototype vehicles and they also have a fleet of six Lexus RX450h SUVs which were modified to drive themselves. By 2012, it was estimated that Google had spent more than 300 man years creating its driverless cars and in August 2015 alone, their fleet drove 70 000 miles autonomously without human intervention whatsoever!
How Far Are We From Fully-Autonomous Vehicles?
While you’ve probably heard about all the hype surrounding the autonomous car, there’s still a bit of confusion about how close we are to seeing these cars on the road.
The truth is that we’re not at full autonomy just yet. And although the technology is getting better every day, it will take quite a bit of time before they’re widely available.
Google has been testing its self-driving car in select cities such as Mountain View and Austin, but there are still limitations on where these cars can travel, as Google keeps them around 35 miles per hour and only drives in sunny weather. While this may seem like an arbitrary restriction, it highlights one of many challenges facing autonomous driving: unpredictable situations require human intervention.
There’s also no self-driving car currently on the market. In fact, some automakers contend that fully autonomous vehicles aren’t likely to be available for another 10+ years.
There are a whole lot of steps between where we are today and complete autonomous mobility for vehicles.
10 Steps to Complete Autonomous Mobility
Not all of the developments are yet in place, but there will be. The road to autonomous mobility will take more than an SUV with some cameras and sensors. In fact, it could take more than a decade or more before we reach autonomous vehicles on our roads. Technologists need to create networked cars, safe algorithms for self-driving cars and technologies to manage driverless vehicles. If that sounds abstract, don’t worry—we’re going to walk you through the full 10-step list below from creation of a working infrastructure (including data collection) onward through regulatory frameworks and legal implications.