Is the flying car merely a sci-fi cliché, or is it a viable transportation tool for the future? One way to find out is to develop talents to drive the concept to market — and that’s what Udacity is doing with its new Flying Car Nanodegree Program.
Online education platform Udacity thrilled flying car enthusiasts with its announcement of the world’s first flying car program. Applications are open until February 7, and the course will start at the end of February.
There are two terms: the fundamental-level Aerial Robotics and the advanced-level Intelligent Air Systems. Each term costs US$1,200 and runs for three months, with approximately 15 hours of weekly study.
Udacity has specific prerequisites for beginners — admission requires good programming skills and basic math and physics skills. Graduates of the Udacity Robotics Software Engineer, Self-Driving Car Engineer, or Intro to Self-Driving Cars Nanodegree programs automatically qualify.
Udacity has secured computer science and aerospace gurus as instructors, including Udacity Founder Sebastian Thrun, MIT Professor Nicholas Roy, University of Toronto Professor Angela Schoellig, and Zurich Federal Institute of Technology Professor Raffaello D’Andrea.
Synced sat down with Udacity Flying Car Nanodegree Lead Jake Lussier to delve into the program’s details and the current state of flying car technologies in general.
Synced: Why has Udacity launched the Flying Car Nanodegree Program at this time?
Lussier: I think Udacity‘s higher level mission is to provide lifelong learning, and this is just a way of saying that we try to provide a learning experience that will help you keep your professional skills up-to-date so that you can continue to succeed and grow in your career.
A flying car might sound very far out, but it follows the same model that we already see the existing aviation industry and with the booming drone industry. There is a great opportunity for students who have skills and autonomy to get jobs now, and we see that there’s a lot of activity right now in flying cars and over the next few years it’s really going to ramp up. So students who learn this stuff now are going to be really sought after.

Synced: What skillset can students expect to acquire?
Lussier: In essence, we’re preparing our students to be very confident software engineers prepared to work in autonomy, in any part of flight and in flying cars specifically.
In the first term, Aerial Robotics really teaches the fundamentals of what one would need to develop any kind of robot, but we offer this specialization in the sky. There is planning, controls, and estimation. The simulation will be a drone in the sky and students will have the opportunity to port their code to an actual drone in flight.
In the second term, we go more into the flying cars specifics, and preparing for a world in which there will be a lot more of these vehicles. We will go into fixed wing flight, and we’ll also talk about hybrid designs. Flying car design has not really converged on one design and has elements of quadcopter and elements of fixed-wing vertical-take-off-and-landing (VTOL) aircraft. We first cover fixed wing designs and how this relates to what they learned in the first term. Then we’ll go a bit more into aircraft, how do the control problems change, how is stability different, and then we get into autonomous fleets.

Then it gets into optimizing missions. If you want to deliver things at multiple places, you have to learn how to optimize to do that effectively. And then when you have entire fleets, how do you coordinate them all so that they operate efficiently, and everything works safely with any central authorities that they’re talking to. So that’s kind of the overarching trajectory of the curriculum.
At every stage the student will not just be learning in the classroom, they’ll also be coding in the classroom, writing Python and getting immediate feedback, and then translating that code into C++ so it goes from code that works well just for understanding the concept, to code that is aircraft ready. Then you can take that C++ code and that’s what you would actually potentially put on a drone.
Overall, they will have an understanding of all these concepts as well as competency implementing these concepts. For the existing industry today, they could bring their skills and autonomy. They’ll be positioned to really be leaders in that space.
Synced: What kinds of companies should program graduates be looking for?
Lussier: At the highest level, these students are going to be very skilled, with experience in building autonomous systems that are not only intelligent but also extremely reliable and robust. If you’re doing image classification on the web, you will be familiar with training algorithms to just predict very well. If you get it very wrong. it’s not a huge deal. But in flight it’s a huge deal if you get it wrong. So it’s not just a simple algorithm but you have to have a system point of view. All of our graduates will have that skill, which is pretty widely applicable.
I think they can find places even in the financial industry where you have to really understand your predictor at a very high level. They can find a lot of roles with those skills. But then in flight specific, most planes fly autonomously for most of their time in this sky. It is really only take off and landing where a human is heavily involved. There are obviously a lot of fully autonomous planes. There’s already a huge market of companies that could find use for these skills.
Then there are the smaller but rapidly growing markets, where they’ll have an opportunity to get leadership positions and really build out those systems. Drones is already a booming industry that was growing very rapidly, and the flying car industry although a bit smaller right now is going to have a similar kind of growth pattern.

Synced: What are the course prerequisites?
Lussier: At a high level it requires a pretty good level of experience in some programming language. You need to be comfortable coding, preferably in Python, C++. You should also have a basic understanding of a bit more advanced mathematics, for example linear algebra probability statistics. A bit about physics since we’ll cover aerodynamics slightly.
If you have not graduated from other programs but you have those skills, then in our admission process, you can explain your qualifications and we will review every application individually and just make sure that you have the skills necessary.
Synced: Could you talk about existing players in the industry?
Lussier: There are a number of startups. For example, Kitty Hawk, which was founded by [Udacity Founder] Sebastian Thrun; Volocopter, which is testing autonomous air taxis in Dubai right now. There is Lilium out of Germany, Ehang in China, and Terrafugia is another one. The Aurora Flight Sciences, Joby. There’s more, this is not an exhaustive list.
Also a lot of the larger companies are either developing these internally themselves or they are developing an acquisition. Aurora Flight Sciences for example was acquired by Boeing. Airbus acquired Bombardier Jet, and Volvo acquired Terrafugia.

There is a growing space of service providers in that industry. For example Iris Automation raised a Series A and they do collision avoidance using computer vision and AI.
UBER Elevate is Uber’s effort to develop a system of users who want to fly from point A to point B. Any flying car maker can enter, and Uber will provide the passengers.
So there are the flying car makers, component developers, and user platforms out there.
Synced: What challenges remain for the technology?
Lussier: One surprising thing is that the basic aerodynamics and the hardware components are largely there. The low level software is mostly there. In some ways, they’re even easier than self driving cars because they don’t have to deal with all the obstacles and people jumping out and randomness. I think this is something that gives us confidence that this actually won’t take that long to get out there because the technology is largely there.
The remaining challenge is engineers who are competent at the intersection of aerospace in computer science, which is exactly why we’re offering this course right now. We traditionally have a pool of aerospace engineers who are not very comfortable with computation, and are used to a slower iteration on products. And we have computer scientists who are more used to running machine learning on the web and not as comfortable running on real world robotic systems.
Once you have those engineers who are fluent in both, there are additional challenges in terms of not just getting a single vehicle to fly stably but to have a whole host of vehicles going 120 miles per hour in the sky. This kind of higher level systems thinking and system optimization is still a challenge today.
Synced: How does flying car technology relate to machine learning models?
Lussier: Machine learning can be used in most of the process. A lot of flight does not permit black box, therefore we can’t just get some training set, train a model and then hope that the performance is sufficiently good. On lower level control problems, we require that we have a physical model and that the performance of the vehicle abides by that model.
But then, a lot of these models have parameters so we can measure the efficacy of the vehicle by its desired path. Whenever you have that kind of setup where you have parameters, this always allows for machine learning where you can optimize those parameters over time based on your measure of success.

Once we move beyond that very low level stuff, when we look at the entire system, this is an optimization problem in the sky. You have a set of resources, and you have some objective that you want to maximize and then you use your data to try to minimize some error function. All levels of this flying car stack, we can use machine learning and AI to try to do a lot better than we might do just using simple programs in simple physical models.
Synced: Apart from technology, what are other challenges for democratizing flying cars, for example, regulations, architecture, battery?
Lussier: Safety is the utmost concern and will really impact public perception of this technology. It needs to be implemented in a way that is safe and also has a super strong track record. So what’s important for the industry is to start having use cases where we’re delivering real value to customers. When we show that value, people will appreciate as we do that it’s safe, and they will gain trust in it.
The other challenges you speak to are definitely when you’re getting things up into the sky. It’s important to do so efficiently and if it’s a really heavy load then it’s going to require a lot of energy and battery, you would have to take that into consideration. That will be more a matter of battery technology progress tackling those problems where the value to the customer can be met by a solution that leverages the available battery technology.
Synced: Where is the flying car industry today, and where will it be in three-five years?
Lussier: An analogous situation would be that of the the car industry where there is a very mature car market increasingly using intelligence. If you’re a career-seeking engineer, there is a big market of car companies right now, and the self-driving car is high-growth.
With flying cars, we have a similar setup where there is an established aerospace market, and there is a smaller but rapidly growing drone and flying car industry.

The self driving car five years ago was in the space where the technology was starting to be taken more seriously but it was a very niche industry. But it was high growth. The flying car right now is in a similar kind of position in that we’re in the early days of it but it’s going to be high growth. The flying car has this nice advantage over the self driving car at that time, because the self driving car has carved out the path technically and socially for a lot of the progress that the flying car will have to make in terms of the AI, navigation, mapping, and coordination.
So, in the next one or two years, there’s going to be a huge amount of growth, and we will see a lot more mass adoption over the next five years.
Journalist: Tony Peng | Editor: Michael Sarazen
Useless course no support no explanation for concepts or questions asked. not worth even one penny