Why GoTo
Eli Osherovich, the Head of Yandex Self-Driving Cars Lab in Israel

Eli Osherovich, the Head of Yandex Self-Driving Cars Lab in Israel: "My daughter will not need a driving license

GoTo interviews smart mobility leaders so that we can learn from each other, share our news and industry concerns.
By Katya Rozenoer
Eli Osherovich
The Head of Yandex Self-Driving Cars Lab in Israel
If you would search for an Israeli-based technical leader with an experience working for Intel, Google and Amazon, someone who has built development teams from scratch, someone who has developed many innovative products millions of people use today, you would find Eli Osherovich. At least this is what Yandex's search engine returned. Recently Eli has joined a newly established Yandex's Self-Driving Cars Lab in Tel Aviv, Israel. Eli's mission now is to build an engineering team that would facilitate the development and deployment of the autonomous cars worldwide.

You have recently moved from e-commerce to mobility, could you describe the transition?

First of all there is this exciting and accidental name matching – I moved from Amazon Go to Yandex.Go. In Amazon Go, I worked on a cashier-less store and here at Yandex we are working on a driver-less car. Very similar concepts in a way.

But do you feel the karmic difference? Commerce in the end of the day is all about making people consume more and mobility is simply about moving them.

This is not how I see things. I do not think that Amazon forces people to consume. Both Amazon and Yandex help people simplify their lives – spend less time doing shopping or commuting, thus letting them spend more time doing things they love. In case of autonomous cars we are also making the roads safer.

Where is Yandex with the autonomous cars exactly now?

Starting 2018 we have our robotaxis deployed in two locations in Russia – Skolkovo and Innopolis, due to regulations there is still an engineer sitting in the car (not in the driver's seat however), but they do not interfere into the driving.
The car that we develop can see much further and much better than a human
This is pretty impressive. I understand that autonomous car development is very expensive, could you talk about the funding situation at Yandex?

We spend what is necessary. Yandex is unique in that sense, I do not think there is another company in the autonomous domain that managed to achieve as much as Yandex in such a short time. Yandex started this effort in early 2017 and we already have robotaxis in the streets. This is very efficient, given that the core development team was just around 100 people when the first results were delivered, I do not know any other company that managed to achieve the same in such a short time with such a small team. Now, with the fleet expansion the team is growing too.

You worked for Google and Amazon before, so obviously you know what you're talking about. What's the Yandex secret?

In Amazon, there are leadership principles and one of them is "being right a lot", meaning that it is expected that the leadership will be able to make right decisions even in the absence of full information and certainty. I think that this applies to the Yandex leadership team, which made a lot of right decisions and managed to avoid many mistakes, including those made by other players in the market.

Could you give an example of such a decision?

An autonomous car needs to "see" things very clearly and there are companies in the industry that are betting exclusively on cameras-based vision. But we have seen many examples of how cameras can be fooled and misled in certain situations. At Yandex, in addition to cameras we also use other types of sensors, so we do not have to deal with these problems.

There is a famous "trolley dilemma" in which a driver is forced to decide who to crush into and kill. How do your algorithms deal with this moral conundrum?

Have you ever encountered such a dilemma in your driving life?


This is because the chances for such thing to happen are quite low. The car that we develop can see much further and much better than a human, so the chances for an autonomous car to get into this type of situation are even lower, the car will prevent it from happening.

But you should have some algorithm that would have to prioritize something in such a situation. What will it prioritize?

We focus on building algorithms that prioritize not ever getting into such a situation. And we make sure our cars see very clearly for a very long distance.

A trolley dilemma-based scenario developed as a part of the MIT Moral Machine project
Ok, but how do you make sure the car sees things? Even if you have a camera with a very high resolution, what about the rain, snow, dirt?

This is actually a really challenging problem. We use multiple types of sensors, not only cameras but also lidars, laser sensors which illuminate objects and then measure the reflected light. By fusing all the sensors' data regardless of the weather conditions and time of day – rain, snow, day, night, we see a complete 360-degree picture all the time.

So should a camera fail, you have other sensors to rely on.

Yes, and safety is our first priority, we constantly monitor all the sensors, if there is a problem and something fails, we safely stop the car.

What is your plan for dealing with zero-day security vulnerabilities that will be inevitably discovered in your software?

First, all the calculations happen on board of the car. The software components responsible for navigation and driving are not contacted to any outside online service, which by default makes it impossible to take control over the vehicle without direct intervention to hardware. Moreover, Yandex has more than 20 years experience of successfully providing information security for company's multiple services (mailing, online payment etc). Our self-driving systems are regularly tested by Yandex cyber security team to make sure they meet company's high standards.

What's the hardest and yet unsolved problem for your Lab in Israel right now?

The biggest challenge for us now is to hire the best people. Once we've done that, we are going to start solving very complex machine learning problems. We will be responsible for one of the very important areas of the autonomous car technology, I cannot specify which one exactly, but we will build a large scale machine learning system, which will ultimately improve our machine learning processes and will help us get better, faster and more robust results.
What is the profile of people you're looking for?

We need three kinds of people:

· Software developers working with big data and distributed computing;

· Researchers or applied scientists, people who can develop state of the art machine learning models;

· And we also need the third type of people which are machine learning engineers who can do both, programming and machine learning.

We are looking to create a mixed team consisting of very experienced engineers and researchers but also from the graduates from the top Israeli universities. In the first year we aim to have a team of 10 people.

Should they be passionate about building robots?

We are looking for people who are really excited about self-driving cars, but who have a passion not in building robots, but in making other people's lives better.

What is your vision for self-driving technology implementation five years from now?

I think we will see a lot of autonomous cars and it will mainly be robotaxis as a part of carsharing services. From then on, I'm now sure how it will go, but I do not really believe in car ownership. Also, I'm confident that my daughter who is now 7 years old will not need a driving license.

What do you think is the biggest industry problem right now?

Technology-wise we are ok, and we should be able to bring it to the level suitable for large commercial deployments. We just need to facilitate regulations and work on educating the public. The idea of self-driving vehicles needs to be explained, the idea of abandoning private cars and adopting carsharing models needs to be explained as well.
You can reach out to Yandex Autonomous Lab in Israel at bd-sdc@yandex-team.com.

Apply for a job at hr-sdc@yandex-team.com

And come see the technology in action at the next Smart Mobility Summit on October 28-28 in Tel Aviv!
GoTo Mobility newsletter
Subscribe to receive our amazing monthly newsletter
By subscribing to the newsletter you are agreeing to our Privacy Policy