Elon Musk: Tesla’s fleet training is orders of magnitude more than everyone else combined

Elon Musk made some bold statements about Tesla’s effort to develop self-driving technology during the automaker’s latest quarterly results.

During a conference call following Tesla’s Q1 2020 results, Musk discussed Tesla’s latest Autopilot development with the new “Traffic Light and Stop Sign Control” feature.

It’s the first step toward Tesla’s system being able to automatically handle intersections — one of the most difficult parts of driving.

As we previously reported, the new feature is not exceptionally useful to Tesla drivers at this point since it requires confirmation to move forward at a green light or after a stop, and it doesn’t execute turns.

However, Tesla is using it to train its Autopilot system to improve and evolve into something that will automatically navigate intersections.

Musk explained that drivers are helping “label” intersections and training its fleet learning system to make it better:

Essentially, the driver when driving and taking action is effectively labeling — labeling reality — as they drive and [make] them better and better. I think this is an advantage that no one else has, and we’re quite literally orders of magnitude more than everyone else combined.

While other automakers and technology companies developing self-driving systems are relying on test fleets and simulation to collect data, Tesla is leveraging its fleet of now over 1 million cars to collect extremely valuable real-world data.

As for the new feature, Musk said during the call that Tesla is already collecting data at a rate of 1 million intersections per month and it is expected to increase rapidly:

We are collecting data from over 1 million intersections every month at this point. This number will grow exponentially as more people get the update and as more people start driving again. Soon, we will be collecting data from over 1 billion intersections per month.

Musk compared Tesla’s advantage to Google:

I think this is difficult to fully appreciate. The reason I say it’s very difficult to have a search engine that competes with Google is that everyone is training Google all the time with their searches. So, when you’re searching something and you click on a link, you’re training Google every time you do that. It’s very difficult for any new search engine to compete on that basis.

The CEO plans to use the data from these millions and soon billions of intersections to improve the system enough that it won’t require confirmation for the driver to go through the intersection and even make turns.

Once that’s possible, it will lead to Tesla releasing its “Autosteer on city streets” and bring it under its “Navigate on Autopilot” feature.

In other words, drivers will be able to enter a location in the navigation system and the car will be able to drive there automatically. However, it will still require the driver’s attention at all times and be ready to take over.

Musk added:

I feel extremely confident that it would be possible to do a drive from your home to your office most of the time with no interventions by the end of the year.

He is basically talking about supervised autonomous driving.

At that point, Tesla will continue to collect data and improve its system through fleet learning in order to make several times safer than human drivers and to eventually convince regulators to enable it to launch its full self-driving system.

During the conference call, Musk said that he hopes that will happen in some markets next year.