Self-Driving Cars - Application of ML

How do self-driving cars make decisions?

The development of self-driving cars is one of the most trendy directions in AI and machine learning. In 2020, we saw advancements from companies like Waymo that allow customers to hail self-driving taxis, a service called Waymo One. Alibaba’s AutoX launched a fleet of fully automated cars in Shenzhen without accompanying safety drivers.  Automotive Artificial Intelligence is rapidly displacing human drivers by enabling self-driving cars that use sensors to gather data about their surroundings. But how do self-driving cars interpret that data?

Self-driving cars can identify objects, interpret situations and make decisions based on observations and object detection. The main part lies in object detection; we use different object classification algorithms. So training the model with object classification algorithms needs lots of data in pictures. The more data we train in reinforcement learning, the better accuracy we get. 
 

“It turns out that reinforcement learning is a type of machine learning whose hunger for data is even greater than supervised learning. It is really difficult to get enough data for reinforcement learning algorithms.” — Andrew Ng.

Here are some common machine learning algorithms used in self-driving cars.

AdaBoost:

AdaBoost is a popular decision matrix algorithm that ensures the adaptive boosting of learned. It combines and adapts the performance of multiple algorithms, so they work together and complement each other. If one algorithm performs poorly, its combined performance can contribute to better performance.

This picture describes the data classification using AdaBoost. It allows for more accurate decision making and object detection in self-driving cars and is especially useful for face detection, pedestrian and vehicle detection.

TextonBoost:

TextonBoost also combines the weak learning algorithm to produce a strong algorithm. It boosts image recognition based on labeling of textons. Textons are clusters(groups of data) of visual data that have the same characteristics and respond to filters differently.
 

The TextonBoost algorithm brings together information from three sources: appearance, shape, and context in self-driving cars. This algorithm combines several classifiers to produce the most accurate results. It looks at the image as a whole and captures its characteristics concerning each other.

How does a self-driving car see?

The three major sensors used by self-driving cars work together as the human eyes and brain. These sensors are cameras, radar, and lidar. Together, they give the car a clear view of its environment. They help the car identify the location, speed, and 3D shapes of objects close to it. Additionally, self-driving cars are now being built with inertial measurement units that monitor and control several acceleration and locations.

Reliable cameras

Self-driving cars have several improved cameras at every angle for a perfect view of their surroundings. While some cameras have a broader field of view of about 120 degrees, others have a narrower view for long-distance vision. Fish-eye cameras provide extensive visuals for parking purposes.

Radar detectors

Radar detectors augment the efforts of camera sensors at night or whenever visibility is poor. They send pulses of radio waves to locate an object and send back signals about the speed and location of that object.

Laser focus

Lidar sensors calculate distance through pulsed lasers locations wearing driverless cars with 3D visuals of their surroundings, adding richer information about shape and depth.

Source

Future Works:

Applied in a self-driving car, machine learning is a powerful technology. Self-driving cars using machine learning will define the future of the transportation industry. And it’s no secret that they’re a perfect match. Machine learning algorithms are most commonly used in autonomous vehicles for perception and decision-making. But there are many more algorithms and possibilities to discover for self-driving cars using machine learning. For instance, you can even apply machine learning to autonomous navigation and a driver’s state recognition.

Today, self-driving cars can already do plenty of things using machine learning. And they’ll be capable of even more in the future. So when vehicles become fully autonomous, you’ll know exactly what drove the change.

Frequently Asked Questions:

1. Are self-driving cars legal?
   In the United States is it strictly illegal to own or operate a self-driving car. Many states have passed laws regulating or authorizing the use of autonomous vehicles to prepare for the changes that self-driving cars may bring. But no state has outright banned the technology.
    
2. Are self-driving cars safe?
   The safety benefits of automated vehicles are paramount. Automated vehicles' potential to save lives and reduce injuries is rooted in one critical and tragic fact: 94% of serious crashes are due to human error.
    
3. Who is responsible if self-driving crashes?
   In fatal accidents involving supervised autonomy systems, U.S. regulators and safety investigators have repeatedly placed blame on human drivers who weren't watching the road. When truly driverless cars hit the road, responsibility will shift from drivers to vehicle makers and software designers.
    
4. Can self-driving cars hacked?
   A new report by the European Union Agency for Cybersecurity (ENISA) finds that self-driving vehicles are vulnerable to hacking because of the advanced computers they contain. The hacks could be dangerous for passengers, pedestrians, and other people on the road.

Key Takeaways:

In this blog we discussed:

• Reinforcement learning
• Algorithms related to self-driving cars
• Hardware components of self-driving cars
• Future works

If you are interested in machine learning take this awesome course from CodingNinjas.