Add to that the fact that you are never going to get 100% fleet utilisation, will need to pay for qualified tele-operators, maintain/repair vehicle along with sensors...
I recently listened to a podcast episode of the Autonocast[1], where they interviewed a Harvard Researcher who claimed the economics of Robotaxis just don't work. Very interesting listen.
For those interested in mobility (including micro-mobility) and self-driving vehicle technology (from the perspective of a VC), I highly recommend Riley Brennan's "Trucks | Future Of Transportation" weekly newsletter: http://www.tinyletter.com/transportation
This looks great! Thanks for sharing. Interestingly enough, from looking at the table of contents, it seems this book starts with a more (and welcome) pragmatic approach, where you write some python code before, look at data visualisation techniques, etc, before delving into stats.
I haven't done the course yet, I've just found it. But, from the rationale video, the course seems to be more about weaving recurrent fundamental data science concepts throughout, emphasizing one particular concept or technique in each chapter, so I guess that it would make more sense to take it as a whole.
It is intended as a "glue" course, having completed CS fundamentals and before core data science courses, like statistics, machine learning and databases, giving students a context for what lies ahead, and just enough to be dangerous and start doing data science stuff.
If this is what you are after, you may also want to consider CMU's "Practical Data Science", which seems to have a similar approach, videos, much more machine learning and big data, and is also very current, but doesn't have such a nice companion online book (but the notes look great) and has much less statistics: http://datasciencecourse.org
Both look like great DS intro courses from top universities, we are spoilt.
And then, also from Berkeley, there is "Data 8", which is intended for those who want an intro to data science, but don't have any programming or college math knowledge yet; it also has a similar online book with working links to Jupyter notebooks: http://data8.org/sp19/ (and videos: https://www.youtube.com/playlist?list=PLXbeRfilLvMoC3QZKxRrp...)
I completed the self-driving car nanodegree last year and currently doing the computer vision one. I did not seek a job at the end of the SDC one but I learned a lot and that could be helpful in the near future, either for my startup or if I decide to look for a new job.
YOLO is a very good and approachable object detection technique. I recently re-read the paper for the original YOLO [1] from 2015 and loved the apparent simplicity of this technique.
Super awesome! I had a similar idea about a year ago but never got to implement it. We need more visual and straightforward ways to create neural network architectures.
Can you currently create the architecture and run a given dataset against it? That would be very cool...
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.
I am currently going through the self-driving car engineer nano degree and I completely disagree. The content was not useful for anything besides trying to to make students be able to complete the projects. They completely skip over and hard theory or mathematics probably because most people would struggle with it
In all fairness there is a big difference between research AI and applied AI/ML. You can spend lots of time on the former before even touching a keyboard. But people can do useful things with the latter with a fairly light dose of theory.
I did the first semester of that course. It was ML 101 repeated several times. It lacked any rigor. I hope to god no-one who finishes that course has an important job working on self driving cars. Moveing fast and breaking things shouldn’t apply to driving.
Everyone...let's be honest with ourselves and what we want from these moocs. I don't know about everyone else but what I want is high-quality courses that are online and basically free. Ads are okay and premium features are okay (like live tutoring or a human given lecture). When I say high-quality I want the content to be similar to what you would get from a Harvard, Stanford or other top university in the field in terms of the content. I want to learn the same topics. Don't water it down and make it simple. I want the real deal with the difficulty and all. Once that is accomplished..work on building tools that make these subjects easier to learn.
This is quite a hurdle to build but ultimately this is what we need. We need a free online university that teaches the same material you would learn from a top university. I don't like the watered down MOOCs that udacity, coursera and udemy offer.
I don't know about everyone else but what I want is high-quality courses that are online and basically free
What you want then is EdX. Every course is backed by a real-world institution that people have heard of, and they are free to audit, pay only if you want the certificate and/or to support their nonprofit mission
There are occasional trash courses on EdX though. If I hadn’t already started on LouvainX’s MicroMaster’s in International Law Amnesty International’s course on the law of asylum and refugees would have turned me right off. The intellectual level of that course was round about middle school.
I've had the same issue! People say that EdX is so great and it is equivalent in course quality but it simply isn't. I haven't' seen that. All of the courses are heavily watered down...and I mean significantly watered down like you don't even learn 20% of what you learn from the traditional course in college.
> We need a free online university that teaches the same material you would learn from a top university.
I think this is Khan Academy [0]. It is a great resource that I've been using for quite a while. They even gamify it a little with points and avatars that you can buy. Funny thing is, like a lot of traditional universities, it lags behind on buzz-wordy courses.
Khan Academy is far from a tutorial site. They have entire lesson plans, lectures, reading assignments, tests and a decent and helpful community in the comments. They cover just about all the material from Kindergarten to a Bachelors Degree.
Khan Academy is a better educational environment than I've ever experienced.
I've done the same comparison and there is no real comparison. Khan excels in the k-12 area because most of the schools these days have extremely limited curriculum and generally teach a series of limited topics in each subject they teach. Sure, for that type of stuff Khan is okay. When you get into the university level stuff...it is definitely not what you learn from a BA degree.
I recently listened to a podcast episode of the Autonocast[1], where they interviewed a Harvard Researcher who claimed the economics of Robotaxis just don't work. Very interesting listen.
[1] http://www.autonocast.com/blog/2020/3/11/177-ashley-nunes-on...