Tech News

Simulation City, the virtual world where Waymo cars learn to drive themselves

Pinterest LinkedIn Tumblr

Gray cubes with a blue skinny surface. You are driving down a dark highway, surrounded on all sides by reckless green squares. You try to skip the gray cubes with the bright blue and change from one lane to another. However, the gray cubes maintain their stability under all these difficult conditions.

Simulation City is described in English as Simulation City. This city is an integrated virtual world in which Waymo, a splinter from Google, trains its cars under the most difficult conditions possible, in preparation for launching them on real roads. The gray cubes previously described are one of the company’s automatic semi-trailer trucks, while the green cubes are the rest of the vehicles on the road.

Waymo is one of the most prominent companies in the market for smart vehicles – or autonomous cars – that are trained using more than one simulation program, including the aforementioned “Simulation City” program.

As for the first simulation program, it is the CarCraft program, which the company has been training its vehicles with since 2017, and its cars have – hypothetically – traveled up to 5 million miles in the training program. Simulation City represents the second generation of this type of software. And the company is testing the system.”Waymo DriverThe operating system that will power the vehicles.

Read also: Ford cars tell you if they have been broken into

Why did Waymo launch the City Simulation Program?

The company felt the need to launch this software, which is the second generation of its simulation software. After “gaps” were discovered in the first program, this was stated by the company’s product manager.

It uses its new software to make sure that vehicles, especially trucks, are truly urban. Indeed, Waymo now has real road vehicles in some suburbs of Arizona, as well as states like San Francisco, Mountain View, California, and others.

Currently, the company is relying on truck training with the new program to begin expanding its physical presence in cities. Currently, there are 600 vehicles being tested, of which 300 are in Arizona alone.

What is the significance of the simulation process?

Simulation in a field like this is extremely important. The company helps train cars to handle as many scenarios as possible that the trucks will be exposed to in the city.

The process is also very large scale, which essentially increases the effectiveness of machine learning. This is in addition to the fact that the basis for these vehicles is to maintain security and safety.

dynamic size image 1625719922 409 72475

As Waymo reported, the company was able to simulate 15 One billion Miles of driving through the simulation city. Unlike 15 Million Mel was driven in the real world by trucks!

The simulation city is more advanced than the previous generation, from a computational point of view. The company is able, through this simulation program, to simulate very simple conditions such as simple raindrops as well as the reflection of sunlight on the car windshield and surrounding cars!

In addition to simulations, real (physical) tests and the company’s test results in its past programs, the company also relies on numerous data from the National Highway Traffic Safety Administration, specifically accident data of course, as well as data from the Transportation Research Council.

And, as you can expect, the company uses artificial intelligence and machine learning techniques massively, including data collected through the many sensors carried in the car, which helps create an imaginary picture of everything around it, a technology that the company calls SurfelGAN, as Surfel is an acronym The term “surface element” or surface factor, while GAN stands for “Generative Adversarial Network,” is a network that simulates the aggressive behavior of other drivers.

Have an article/sponsored post to share? Whatsapp: +2348129656985.

Notify of
Inline Feedbacks
View all comments
Pin It