Added a reflection for sem2 2023 uni

_posts/2023-11-29-2023-sem2-reflection.md

@@ -0,0 +1,175 @@
+---
+title: 2023 Semester 2 Study Reflection
+date: 2023-11-29
+categories: [Study]
+tags: [study, engineering, control systems, robotics, dynamics, learning]
+toc: true
+---
+
+## My Goals
+
+So at the beginning of the semester I wrote [this]({ %post_url 2023-07-24-starting-a-blog %}) post
+about starting this blog for uni and the goals I had for the semester. The gist was that I felt I
+was spending too much time taking notes and I wanted to try writing blog posts instead as I thought
+this would help me to learn the content better by having to summarise and explain it.
+
+So, did I follow through with these goals? Well partly... I don't think I took any notes in classes
+this semester but writing technical blog posts was more time consuming than I imagined and it was
+unrealistic to write one for each class each week. Most of the courses I took this semester were
+quite time consuming with assessment items and some amount of team work in each which meant I made
+the decision to focus on completing assessment and going to classes rather than writing more blog
+posts.
+
+I finished my last exam almost two weeks ago now and so more recently I have been writing more blog
+posts on some of the content from some of the classes \(mostly things relating to control
+engineering and robotics\). I aim to continue writing some of these technical posts about the things
+I have learned this semester over the coming weeks.
+
+## Results
+
+So how did my strategy of taking no notes in class go? Well firstly - as a good engineer - I think
+it's important to note some of the limitations of this little experiment. Namely, this is just my
+opinion and experience so it really just serves as an anecdote rather than any _hard_ evidence.
+
+To cut to the chase I got all 7s \(High Distinctions\) for the four subjects I did
+\([Control Engineering 2](https://course-profiles.uq.edu.au/student_section_loader/section_1/128875),
+[Robotics and Automation](https://course-profiles.uq.edu.au/student_section_loader/section_1/130244),
+[Advanced Dynamics and Vibrations](https://course-profiles.uq.edu.au/student_section_loader/section_1/130174)
+and
+[Professional Practice and the Business Environment](https://course-profiles.uq.edu.au/student_section_loader/section_1/128579)\).
+
+And to be honest, it didn't _feel_ like choosing NOT to take any notes impacted my learning very
+much compared to other semesters. It is of course hard to be objective about these things since I
+don't have a perfect control for this experiment. However, I do think I was able to engage in class
+better and understand concepts as they were presented a bit more easily since I could think about
+them rather than trying to write them down.
+
+One other new thing I tried this semester was using the recommended textbook excerpts as my primary
+resource for content for my Control Engineering 2 course. In the past, I have not bothered much with
+textbooks and mostly just learned content from the lecture slides. I have to say though, the
+textbook excerpts that were provided were _very_ helpful. They covered the content thoroughly and in
+many cases to a greater depth than the slides. Yes, sometimes it could be a bit difficult getting
+stuck on a part and not being able to ask questions but usually continuing on and coming back to it
+later was a good strategy.
+
+## Short Course Reviews
+
+### Advanced Dynamics and Vibrations
+
+This course was a classic mechanical course broken into three sections.
+
+The first part delved into multi-degree of freedom linear systems with a focus on modal analysis.
+The assignments focused on a hypothetical consulting problem for a wind turbine where passive
+methods were used to control vibrations in the system. This means designing dimensions for the
+blades or gearbox e.t.c to minimise vibrations during operation.
+
+The second part moved into continuous systems focusing on strings and beams. Again, modal analysis
+was used to analyse torsional, longitudinal and transverse vibrations. Furthermore, some approximate
+techniques \(Dunkerley's and Rayleigh's methods\) were used to model the effect of point masses.
+Finally, there was some brief dicussion on Finite Element Analysis \(FEA\) which uses software to
+model continuous systems by breaking them down into lots of small parts making them MDOFs. We used
+ANSYS software for part of the consulting assignment to model the effect of tapering turbine blades.
+
+Finally, there was a brief discussion about instruments used to practically collect vibration data
+such as accelerometers and how to analyse this data. The effects of noise and discretisation of data
+on the spectrums produced using FFTs are some examples of this.
+
+Overall, I didn't mind this course. There are parts such as the signal analysis at the end which I
+thought were slightly more interesting but the ANSYS modelling felt a bit tedious and uninteresting.
+
+### Robotics and Automation
+
+This was my favourite subject this semester as I predicted at the beginning. I have heard in the
+past this course was more heavily focused on inverse kinematics for robotics however this year there
+was a new course coordinator and she seems to have made the course much more broad.
+
+The first part of the course mostly focused on stochastic models in robotics for things like
+movement and sensor models, mapping, localisation and SLAM. Next some path planning was introduced
+with some classical graph traversal techniques including Dijkstra's Algorithm and A\* as well as
+techniques for creating these graphs such as PRMs and RRTs. Next, we looked at some forward and
+inverse kinematics using screw coordinates and then the homogeneous transforms associated with
+cameras. Finally, there was a brief overview of how Reinforcement Learning could be used for some
+more difficult tasks in robotics. RL is something I am very keen to learn more about so this was
+cool to see.
+
+The assessment was based on programming some turtlebots using ROS2 \(in Python\) to run SLAM. We had
+to design an algorithm to fully explore a static map \(that we did not know ahead of time\) and
+localise some ArUco markers around the map.
+
+### Control Engineering 2
+
+This course was also pretty interesting and introduced more control techniques.
+
+The course actually began with nonlinear control using Lyapunov theory which I think was kind of a
+strange choice but nevertheless I thought the concepts were quite interesting and this was probably
+my overall favourite part of the course. Next, linear state space models were explored including
+many of the concepts I have been writing blog posts about recently such as [state feedback and
+controllabliity]({% post_url 2023-11-21-controllability-and-stabilisability %}), [reference
+tracking]({% post_url 2023-11-26-reference-input-tracking %}),
+[LQR/I]({% post_url 2023-11-27-lqr %}), observability and Kalman Filters.
+
+Overall, I thought this course was pretty good but I think some more depth or even other techniques
+for control such as RL would have been cool to see.
+
+### Professional Practice and the Business Environment
+
+Although I didn't mind the idea and content of this course it just seemed like it was missing a lot
+of depth. The main focus was on business management with assessment based around a startup idea in
+groups. The course began with an overview of how to come up with a value proposition and then
+continued with different analyses that could be conducted for the proposal. First the Discounted
+Cashflow Analysis was used to help get an idea for the value of the business cashflows. Next there
+were some details about what it means to be a professional engineer and how to get chartered after
+graduating. Finally, there were some details about business sustainability.
+
+Most of the content was very easy and lacked depth. I really wish there was a bit more to it because
+I think I could have really liked this subject if that were the case.
+
+## What Next?
+
+Well in the next couple of months I have a few things I will be focusing on.
+
+Firstly, I will hopefully continue to add some posts about the things I learned in the last semester
+of uni to help consolidate my knowledge, have some personal notes to look back on and to hopefully
+help anyone else who might stumble upon them. It would be cool to look into adding some of these
+posts to a more public blogging website so that could be worth a look too.
+
+I also want to lean in more to software engineering and particularly AI since this is what I am most
+interested in at the moment. For now, I have been focusing more on some fundamentals and trying to
+complete the daily leetcode challenges mostly in C++ to keep up my skills in that language. Aside
+from that, I plan on doing the [Advent of Code](https://adventofcode.com/) challenges this year
+starting on the 1st of December in C++ or maybe Python or even TypeScript. Again, I just want to
+keep my skills up in these languages and this seems like a good fun way to do so similar to
+leetcode.
+
+I'm also hoping to finish off an MVP for an app for a tutoring business I co-founded. It's quite
+small at the moment but I am hoping that getting this app done that helps automate some of the admin
+will mean we can start to scale up soon.
+
+After that I will be starting my placement at [Hatch](https://www.hatch.com/) in mid January where I
+will be commencing a six month project for my Master's thesis. The project will be based on computer
+vision which is great because as I said before I am keen to learn more about AI and deep learning.
+
+In the longer term, I will have to start thinking about what I want to do after I finish hopefully
+at the end of next year. The safest route most people take is to just apply for a bunch of grad
+positions and work that for at least a couple of years and I will definitely be on the lookout for
+applications soon but there are a couple of other options I am also seriously considering.
+
+Firstly, I do think that I will be interested in further study and probably a PhD at some point and
+so I may start applying for those opportunities as well. I'd really like to research something in
+AI, IoT or biomedical engineering that could benefit disadvantaged people. For example, I think
+there are some really exciting developments happening in AI right now related to logical reasoning
+and chain-of-thought and I would love to look at how these could be applied to education systems.
+From tutoring I have learned that helping students to learn is all about guiding them by prompting
+them with the right question which relies on gauging what they do an don't understand. In other
+words, its important for you and the student to understand what steps in they struggle with in
+problem solving and focusing on that. If AI was capable of doing this I think this support could
+become affordable for so many more students and improve educations worldwide. Regardless, I want to
+start looking for papers in these areas and see where might be good to apply.
+
+Another option I am considering is looking at a startup. This is the most scary because I feel like
+there are so many things I don't know both technically and in terms of business and management. It
+also seems obvious that my initial income would be much worse and if things do not go well I am not
+sure how convincing employers of my skills would go. On the other hand, it seems like being young is
+the ideal time to take these kinds of risks because I don't have much responsibility to other people
+that I need to worry about. Regardless, at some point in my life this is something I would like to
+do since I believe it is one of the most impactful things I could do.