Self-Driving Car: Latest Technologies (Revised)

The Dream Has Become a Reality

Self-driving cars might seem like an imaginary machine for people who are not very familiar with the term, but they are widespread soon. According to Grand View Research, the global self‑driving cars’ market size is expected to expand at a compound annual growth rate of 63.1% from 2021 to 2030. They are a key innovation and have high growth potential in the automobile industry.

Weymo

While the market of self-driving cars is expanding, there is one company that attracts the most attention to itself now in the automobile industry, which is Weymo. Weymo itself does not produce cars, but it develops AI software for self-driving cars. Their autonomous driving technology is said to be one of the most advanced in the world.

Image: Grendelkhan/Wikimedia

Machine Learning

Waymo is using machine learning in many ways to improve its autonomous driving technology. Machine learning is a method of having a computer read data and analyze it based on an algorithm. By iteratively making a computer to learn the data of a particular case, they can discover the features and patterns in those data. Then, they apply the found features and patterns to analyze the new data.

Image: Avimanyu786/Wikimedia

Utilization and Training Of AI by Waymo

Self-driving cars need many data in various situations, but it is difficult to train them using real-world data in high-risk but infrequent situations, such as when pedestrians jump out from behind a stopped car or in a snowstorm. Thus, the cars are instead travelling billions of miles using virtual reality simulations. Waymo’s researchers have revealed that they usually drive about 25,000 cars in simulations to keep improving machine learning algorithms.

Recognition Of Objects and Surroundings

The most basic task of self-driving cars is to recognize surrounding objects. Waymo uses a neural network that imitates the mechanism of the human brain (it is essentially a machine learning model) to accurately detect traffic lights, bicycles, pedestrians, lanes, etc. in any weather condition. Weymo has recently released a video of a car understanding human gestures. The video below shows a self-driving car stopping at a crossing with a broken signal and following a police officer’s hand signal.

Source: Weymo

Predicting the movement of pedestrians and vehicles is essential to avoid accidents. In 2019, Waymo patented a system that sends data from a car sensor to a neural network to predict the position of a nearby car.

Future Prospects of Self-Driving Cars and Issues

Though the technologies of self-driving cars are rapidly developing, there are still many issues that cannot be solved by technology alone. For example, in regards to Tesla car fatal accident that occurred in 2108, issues such as who should take responsibility and driver’s moral hazard have been pointed out. It won’t be long before self-driving cars will be part of people’s daily lives if these problems are solved.

 

ーShunya Sunami

4 responses to “Self-Driving Car: Latest Technologies (Revised)

Leave a Reply