Leo Orenstein, Senior Controls Engineer: “If you want to do something challenging, go for autonomous trucking.”

This month’s employee spotlight is on our Senior Software Engineer, Leo Orenstein, who is designing control code for Starsky Robotics trucks. As the controller is a safety-critical multibody vehicle weighing over 40 tons and over 20 meters long that is supposed to operate autonomously on a public highway, there is no doubt it’s a hard problem to solve. Leo says he is enjoying every single part of it and looking for more people who are not afraid of challenges to join the team.

Leo, let’s talk first about your role at Starsky. What are you and your team working on?

I’m on the Planning and Controls team at Starsky. What we are doing is taking very high-level context behaviors such as “keep driving”, “change lanes” or “pull over because something has happened incorrectly” and turning them into specific commands that a self-driving truck can actually follow. The output will be “turn the steering wheel 13.48 degrees right” or “press the throttle 18.9 degrees”.

In other words, we take these pretty abstract ideas and translate them into a language that our computer hardware system can understand and follow. It’s a two-step process. It starts with Planning to identify these abstract things and break them down into tasks that are more detailed but still not quite reconcilable. Then Controls helps turn them into real commands.

I’ve been doing both Planning and Controls, bouncing between them, depending on what’s more critical at the time. Right now, I’ve been working more on the path planning side, and I find it incredibly interesting. It’s a relatively new field as opposed to Controls which is pretty well-established as it has existed for about 70 years now. Path planning has more liberty and is more open for experimentation.

How big is your team now?

There are six people on the Planning and Controls team at the moment, and we are hoping to recruit another team member by the end of the year.

You have hands-on experience of working in many different industries, including Oil and Gas, Transportation, Mining, and Aviation. What was the most interesting job you had before Starsky?

I was working at General Electric’s research center and that was a really interesting job because it was very diverse, and that was where I gained experience in so many different fields. There was this thing that we used to say to each other back then: “If you don’t like what you’re working on, don’t worry. It’s going to change soon.”

It did change a lot. For example, in the same month, I went to an offshore oil rig, a sugar cane plant, and an iron ore mining facility, because I was working on all these different projects. It was intense, but I enjoyed that variety. The work itself was interesting enough, but I especially liked working on different subjects, going from one to the other and quickly switching between them. Each project was unique. Industries, companies and their problems were completely different, and every time I managed to find the right solutions for them, it felt great.

As a person who has worked in both large corporations and small start-ups, can you compare these two experiences?

I’m definitely a start-up person. I have little question about this now. I like the agility of a start-up. I know this is a cliché, but it’s true. I believe in the idea of trying things. If you have an idea, try it. If it doesn’t work out, find something else and then try again.

At large corporations, you have cycles. Let’s say, we start working on a project. Three months in, we know it won’t work. However, it has funding for the whole year and it’s in scope. So, even though we know it won’t work, we keep trying because that’s the plan. I find this dreadful.

Of course, start-ups have their own issues too. For instance, whatever was working when there were 10 people in a company is not going to work when there are 20. It’s not going to work again when there are 50, and if a company doesn’t realize that, the issue becomes quite pronounced.

Besides that, it’s not a secret that big companies have more well-established processes. Sometimes it’s enough to just click on a button and have magic happen. Not a lot of magic happens in a start-up. If something is being done, you probably either know who’s doing it or going to be doing it yourself. I like working on lots of different things as this is the only way to actually get to know your product and understand how the magic is made.

“If you have an idea, try it. If it doesn’t work out, find something else and then try again.”

How has Starsky helped you in your professional development, and what advice would you give to prospective Starsky candidates?

Before I joined Starsky, I thought I was a decent coder. Then I figured out I was wrong. From a technical perspective, Starsky is a really great place to learn. The company has a very open, collaborative environment and the best culture for learning. It basically says, “if you don’t know things, that’s okay, let’s find out together.” It’s part of Starsky’s DNA. So, if you are joining the autonomous driving field from another industry, go for Starsky. We understand that no one knows all the answers, and we are willing to work with new people to ramp up our collective knowledge.

That being said, trucks are the hardest control problem I ever faced. It’s a very complex system. Even for human drivers, it’s a difficult thing to operate. There are many external factors affecting it and a lot of things can go wrong, so you need to be very precise. For instance, we can all of a sudden get a gust of crosswind. It’s almost impossible to predict it and quite hard to measure it, and just as sudden as it appeared, it may go away. However the truck cannot allow this to push it to the side. So, you need to figure out a way to overcome all these changes and make sure that the truck still responds well.

What’s great is that this is not a research project. We often say to each other: “Is there a simpler way of getting it done?” That’s because we are building an actual product rather than just trying to find a theoretical solution. So, we are looking for people who care a lot about turning things into reality. If you do care, if you are ready to push the boundaries, and if you want to do something challenging, then go for autonomous trucking.

“We are building an actual product rather than just trying to find a theoretical solution.”

What do you find the most challenging in developing autonomous driving systems?

Safety is the most challenging part. In general, the more well-defined a problem is, the more feasible and easier it is to solve. With a safety-critical system like an autonomous truck operating on a public highway, it’s like trying to solve a problem where absolutely anything can go wrong. So, you have to take a very disciplined safety-engineering approach and make sure you are covering all your bases. You need to find out all the failure cases, document them and implement safety mechanisms for all these scenarios. Even if your algorithm works 99.99 percent of the time, it will still be failing once a day. So, you need to make sure that the whole system is really bulletproof.

Can you share a few interesting facts about yourself to let people know you better?

I like to cook a lot, and I actually went to cooking classes for about a year at the time I was doing my master’s. I was studying, working and doing cooking classes. That was pretty intense. The breaking point was when someone asked me to open a restaurant with them. The guy had a restaurant space and asked me to open a brewery in it. I did the math and decided that it would be too much risk for me, so I passed on that opportunity.

That’s pretty much when I left cooking, as I figured out that I love it as a hobby. My wife tells me that the only thing that can really get me mad is getting something wrong when I’m cooking. I’m a very chill guy, but if I get the recipe wrong, I get crazy mad for the whole day.

Also, on a personal note, I’m having a baby soon. And I really appreciate how supportive of that Starsky has been. Not only do we have parental leave, but people truly understand the importance of that. I know that some companies don’t really care — even though you’re having a baby, you have to deliver a product in the first place. It’s more like taking parental leave but being on Slack while doing it. At Starsky, you are not simply getting the leave, but you are actually encouraged to enjoy it and bond with your family.

***

If you want to join the Starsky team and help us get unmanned trucks on the road, please apply here.