I started taking some online robotics classes. Very interested in learning more about industrial automation and I thought this would be a good foray into the subject.
I started off with the Robotics Micromasters curriculum on EdX (you can take almost all the individual courses for free). I found the classes to be really interesting in terms of the science and theory of how visual perception works, BUT the MATLAB coding and the multiple teacher styles felt really… academic. In the world of robotics you code in C++, MATLAB, or Python. I am a Python and C++ guy… so unfortunately I just wasn’t interested enough to learn both MATLAB and Robotics theory.
To do some more work in Python I moved over to Udacity and applied to their Robotics Nano-degree* and started the Self-Driving car course. The self-driving car is really well taught by Sebatian Thrun. I was really impressed with how pragmatic his approach is. Much less academic and much more “get it done”, which is exactly what I wanted.
I am in lesson 15 of the course now and the teaching style + the fact it is all done in Python where a great fit. Big lesson here was that you have to find the right teacher for you. I was a little sad after PennX, thinking I didn’t have “it”; but as soon as I switched to another teacher/language it was 10x easier for me and I started to fly through the class.
I still need to do another lap through some SLAM materials to fully make sure I understand all the underlying science. I definitely PID control. I have an OK understanding of policy generation (A* and dynamic programming). My biggest weakness is the Karman filters vs particle filters. I understand the theory but don’t have a great sense of when to implement one or the other.
*I was ultimately accepted into the nano-degree but opted not to do it after talking to a few current members. The initial emails were a little too pushy/marketingy and I saw lots of issues inside the slack channel with the curriculum + no real “Project” output in terms of non-simulated robotics. I did the first project which was the NASA Rover Challenge. I thought it was really cool; but also since all of the projects are basically sourced from other groups (NASA, Amazon Robotics Challenge, etc.), I thought I should give those a shot without spending $1200 and not building any actual robots…