#47 overall
Self-Driving Car Engineer Nanodegree
Udacity
11 recommenders · 12 threads · recommended in 5 years, 2016–2023 · score 12.1
A paid Udacity nanodegree program in which students build self-driving car software, covering computer vision, deep learning with TensorFlow and Keras, sensor fusion, path planning and control, with a capstone that runs student code on a real car. Commenters describe it as a hands-on, project-driven program.
Commenters call it the best hands-on course they have taken and say it lets you build real self-driving software, including a TensorFlow project.
Machine Learning & AIintermediatepaidproject-based#self-driving cars#deep learning#tensorflow#computer vision#path planning#udacity
Recommendations by year
20062026
I want Udacity to succeed as a business. The amazing course content of Udacity especially in AI and Machine Learning has no parallels when it comes to actually accomplishing something novel(for the student). If you take self driving car nanodegree, you will at the end of course be able to code actual software for the car.
I took their Self-driving car Nanodegree and it was hands down the best online course I've ever taken in my life (and I took some crazy courses from MIT and Stanford). Where else could you end up with a capstone of actually programming a real self-driving car that is then running your code on some course in Mountain View?
Offerings right now:
1. If you want academic rigor only, go with edX
2. If you want academic rigor and some more approachable/practical courses, go with Coursera; or if you want to do full MS/MBA/MSA online (they have the widest selection right now comparing to edX/Udacity)
3. If you want unique practical skills in hot areas like self-driving/flying cars, deep (reinforcement) learning etc. without fully committing to rigorous practices, go with Udacity
4. If you need anything else, go with Udemy, Masterclass, School of AI and similar
Agreed that Udacity's AI for Robotics was a weak course. I highly recommend some of Udacity's recent paid-only courses on robotics:
Self-driving car engineer nanodegree (not the one with "intro" in its name) https://www.udacity.com/course/self-driving-car-engineer-nan...
Robotics software engineer nanodegree https://www.udacity.com/course/robotics-software-engineer--n...
Flying car and autonomous flight engineer nanodegree https://www.udacity.com/course/flying-car-nanodegree--nd787
Just as a sample, this is one of many projects I completed as part of the self-driving car engineer nanodegree. My code controls a car driving on a highway with other cars.
https://github.com/ericlavigne/CarND-Path-Planning
It depends on the quality of the course and how close it aligns with my goals.
Ex: Udacity's SDC Nanodegree amounts to $2400, but was worth every penny for me. On the other hand, there are numerous machine learning courses on Udemy, and most of them I wouldn't pay for at all (even at $5-$10).
In the latter case, most of the content is either non valuable (its stuff that's repeated all over the internet) or non practical (can talk a lot about theory or math but if I don't actually build anything it doesn't help me).
Yes definitely possible. I did this 6 years ago.
1. Find the best 1,2 courses for AI/ML At the time the Udacity self driving course was a great course from a basic Udacity ML intro course to a full system that
2. Allocate 2-4 hours a day for this. This is a heavy course and a lot to learn so you have to work really hard to get this done.
3. Final project should be impressive to people int he field. So for example I implemented a YOLO alternative from their paper. You’ll have to do something similar and show results.
Then getting a job is a completely different skill and you’ll be looking at a job on the margins. Keep backup options in lower pay jobs in startups or in non tech companies if your dream jobs don’t pan out.
It could take 6 months to 9 months to understand the content and then 6-9 months to get a job. IF you are okay with that … then do it.
Having ML + software eng background is really good spot to be in.
Check out udacity self driving car class.
I took the course in 2015, when it was Matlab-only and Russ was working on a C++ version; it was one of the best courses in robotics I've taken (together with Udacity's Self-driving car ND). It also had a simple version of Boston Dynamic's Atlas robot that was massively improved recently. Does anyone know if the course contains new content related to the recent advances at BD? Is the optimization code now compatible with ROS? Thanks!
This seems like a necessary reset, especially given their very high valuation. I personally enjoyed most of Udacity's Self-Driving Car Engineer Nanodegree. The content was great and at the time I had a very supportive mentor (gutted they got rid of this feature). Moreover, they have managed to build a large community of current students and alumni who keep in touch even after the nanodegree is over.
I would suggest they reduce the number of courses they offer and improve quality. I tend to see them as a more "premium" MOOC but I do agree that quality of content is variable depending on course.
Another area they should naturally look at is training for employees in corporations. I really hope Sebastian Thrun can turn things around.
Hello! Great questions.
1) If you already have a good grasp of Python, I always advise to start with the AI for Robotics MOOC at Udacity, which is my favorite class of all time. Once that's done, I'd take a look at their Deep Learning classes and the Self-Driving Car Nanodegree.
2) I think it's crucial to get to grips with how the whole stack works, so I always advise to get to grips with a middleware like ROS. Also, don't be afraid to dabble in algorithms (think problems in motion planning, computer vision, etc.)
3) The traditional programs (think PhD programs) create a lot of specialists focused on a single domain, but the industry is in dire need of more generalists. An engineer who is able to dive into any part of the stack is a huge value-add!
As ocdtrekkie mentions, the CommaAI is an option that offers a little hand holding in the sense that instructions are available. I actually went through the process of building one and hooking it up to my car and deciphering my cars CAN bus messages. It's pretty interesting and would definitely recommend it.
As for making your own, it depends on how good you want it to be. For example, the CommaAI is simple to do even from scratch but not super cutting edge. On the other hand theres companys out there using additional radars and Lidar which increases the cost and complexity. But in theory you'd just need to mount a machine in the trunk of your car, hook it up to your cars computer (have a car capable of being controlled solely via computer), hook up any additional sensors. Then write the software that can fuse all of the incoming data and use it to send messages to the car.
If this is something you're interested in learning about you can check out Udacity's SDC Nanodegree
I hope that changes in the future! Looking for a really good course on TensorFlow. Udacity's Self-Driving Car nanodegree is also very hands-on in terms of learning TensorFlow. You build your own mini version.
I've done Udacity's ML and doing the SDC in free time.
Short answer: Yes - I would do it again.
ML (Mid 2016): Cost-wise you may be able to gather the similar quality of materials for free, nothing are too unique and in many cases additional intensive research are still required. But, the Udacity provides nice structure and helps you to keep motivated.
SDC (2016): Much better experience for students, the quality of materials are higher, and amount of support from the peer group are extraordinary. Just one of the five projects in the first term are more complex than the whole ML nanodegree.
Haven't tried the AI nanodegree myself, but because it's a part of the new batch of programs I believe the experience would be quite positive (closer to SDC).
If you will decide to proceed with self-learning path, here is a nice multi-month study plan: https://github.com/ZuzooVn/machine-learning-for-software-eng...
In any case I strongly advice to take a look into Andrew Ng's coursera courses and Andrew Karpathy's CS231n on CNN (http://cs231n.stanford.edu/) as a supportive materials.
Also, if money is the main concern, you may want to apply to sponsorship, or maybe discuss this with your current employer.
Feel free to PM me on twitter (same nick) if you have any other questions, would be glad to assist.
I highly recommend the Udacity courses on machine learning. They even have ones on how to become a self driving car engineer.
You do something similar in Udacity's SDC nanodegree but you use their simulator rather than a new car. Interestingly through trial and error on my project, ELU activation functions was the only thing that prevent vanishing gradient problems. You use the same activation function and I'm curious why you selected that one. I always wonder why it was the only function that worked for me.
It's a bit harsh, but not unreasonable. You can buy this level of self-driving at many car dealerships right now. If you take the Udacity self-driving course, you could build this. Doesn't mean you're ready to ship a road-ready product. Remote control helps some, but it won't get you out of trouble. By the time the remote driver is connected and has situational awareness, it's too late.
This isn't webcrap, where it doesn't really matter if it fails now and then. People die when self-driving vehicles screw up. Especially with heavy vehicles. California DMV doesn't allow self-driving heavy vehicles on the road yet, and that's probably a good decision.
If YC is going to fund stuff like this, they need to get some avionics people on the team. Before they kill somebody.
Toyota got into autonomous driving late. Then they hired Gil Pratt from MIT. Pratt used to run the MIT Leg Lab, which Raibert ran before Boston Dynamics, but Pratt didn't do much there.
It's clear that autonomous driving can be made to work, because Google/Waymo is doing it. It's hard and expensive and it takes a lot of sensors. It also takes extensive testing. Waymo drives 25,000 autonomous miles a week. Volvo has level 3 working on some freeways in Sweden; their 100 users are not required to watch the road while in auto mode.
There are other startups trying to do it. There's the "fake it til you make it" crowd - Otto and Cruise. (Otto's highly publicized Budweiser truck delivery demo was on a nearly deserted freeway surrounded by chase cars.) There's the "it's just a small matter of software" Tesla approach. There's the "throw machine learning at vision and hope" crowd, some small startups. That's what you get if you take the Udacity course and start coding. 43 companies have California DMV licenses for autonomous vehicle testing.
Toyota has been making some bad business decisions lately. They don't make battery electric cars. (They're fixing that, but won't be shipping until 2022.[1]) Instead, they've been pushing cars that run on hydrogen. Toyota sells the Mirai in California, and has a few hydrogen stations so it can be refueled. They sell about a hundred cars a month.[2]
[1] https://www.reuters.com/article/us-toyota-electric-cars-idUS... [2] http://carsalesbase.com/us-car-sales-data/toyota/toyota-mira...
I'd be curious to know why myself?
I'm about 2/3rds of the way thru the second term (I'm in the November cohort), and I plan to finish the nanodegree.
What I have found, though, with this second term, is that it seems to be a rehash of the CS373 course (I took that back in 2012, when Udacity first started) - though it also has taught a few new concepts for me (I like the fact that it is mostly done in C++ - which had gotten really rusty for me - but now I am liking C++ more).
I'd also recommend the Coursera ML course - from what I know, it is basically identical (if not the same material) as the original ML Class from 2011 (I took that one, which was my intro to MOOCs). If you can wrap your mind around the programming language used (Octave - basically an open-source clone of MatLab - where the main primitive is a vector/list), you'll find it does a great job at explaining many concepts that you might have had trouble with in the past (when we got to neural networks and backprop, it finally clicked for me - same with how think in terms to figure out how to parallelize processes).
CNN == Convolutional Neural Network
That...is not a good approach.
Don't get me wrong - learning this stuff is hard, and implementing it (on your own - that is, a completely from-scratch not-using-any-libraries neural network) is even more non-trivial. You need to know and understand linear algebra for this; knowing a bit of calculus (derivatives mainly) can also help, but isn't absolutely necessary.
Trying to build it yourself, while admirable, will likely only lead you down paths already traveled and found wanting (if you are lucky). Once you understand the basics of neural networks and other machine-learning tools (gradient descent, state vector machines, etc) - and have implemented them yourself - you'll have a foundation on which to build with tools like TensorFlow, Keras, and others, without them feeling like such "black boxes".
Remember: Abstraction is a good thing. If you are doing any kind of programming that isn't hand-assembly or plugboards - you are using an abstraction to the machine. One thing I have learned in listening to people like Andrew Ng and Sebastian Thrun - is that even they rely and use those black-box abstractions, and don't have a "not-invented-here" attitude. There's nothing wrong with wanting to look under the hood for understanding, but don't let too much hubris get in the way of thinking that you can do it better without understanding what has come before.
You can take this advice or leave it - it's up to you of course; but if you want to have a better base on how all of this works, then my suggestion would be for you to gain an understanding first of linear algebra (just the basics - vectors, matrices, mathematical manipulation and calculation, etc) - Khan Academy or other resources can help there.
Then, go to Udacity, Coursera, and/or some of the other MOOC providers, and find a class or two on machine learning, and take them. Most will take (if seriously studying, and depending on existing work load) about 6-8 weeks - but you can usually go at your own pace (so shorter or longer learning terms are possible). I'm partial to Coursera's and Udacity's courses, mainly because in 2011 I took Andrew Ng's "ML Class" MOOC and Norvig/Thrun's "AI Class" - from both of which Coursera and Udacity (respectively) were founded.
Later, in 2012, I took Udacity's CS373 course (it was titled differently than today's offering, but I highly recommend it regardless). Today, I am enrolled in and progressing through the Udacity Self-Driving Car Engineer Nanodegree (so I am a bit biased, I suppose).
The tough thing about learning this stuff on the internet, without going thru a well-known MOOC or other vetted source, is knowing what you're learning is valid to learn. There is a ton of information out there purporting to teach you about neural networks or other "AI" tools - and they are actually teaching out-of-date technologies and techniques. But without having a grounding in the long history of machine learning (and AI) as a field, you wouldn't know this. It can be very confusing. There have been some good postings here on HN that have given some details on good resources, so you might want to check those out. There's nothing wrong with learning historical methods (in fact, I encourage it - I believe that one can't really understand a subject until they understand the paths that were taken in the past - whether they led to failures or to success for the period). But you don't want to learn something that is actually sub-optimal for the task.
Finally a couple of other points: Do you know what MNIST or ImageNet is? If not, then you need to learn about datasets and where to get them. Also, for a real CNN, look into the history of character recognition using the MNIST dataset - eventually, you will find a guy named "Yan LeCun" - who is one of the original people who led us down this whole CNN and Deep Learning path (along with plenty of other names).
There is so much out there (much more than I can summarize here) - you have the interest, I can see that. Take the time to properly learn this stuff, and you'll be rewarded with some interesting results.