Summer of Math Exposition

I appreciated this video and believe to be the right audience: a good background in mathematics with no prior knowledge of the Kalman filter.

I didn’t give a higher grade because I felt like the video was a bit long, especially towards the end, when several minutes were dedicated to a single table. I liked the rest and found the pacing and the animations quite pleasant and easy to follow.

While there are some good visualizations later in the video the aid your explanation, the majority of the video is very dense in symbols and feels more like an audiobook of a textbook than a textbook.

That being said, this is a veritable tome of tips and tricks for programming with measurements. It’s also clear there was a LOT of thought put into the modeling presented, even though its not as tangible as other submissions I’ve seen.

You make the motivation for the filter clear, and I like that you build up the model to gradually add complexity.

There are a lot of variables in your equations, and you move through the definitions quickly to get to the formulas. It would help to have those definitions visible somewhere so a viewer doesn’t have to keep them all in mind while also learning new content.

The first half of the video is equation-heavy, and you often refer to parts of one equation when there are many on screen. Highlighting the relevant terms would make it much easier to follow.

It took me several minutes to understand what this problem was about. I think you need to introduce what the Kalman filter is much earlier and not assume the viewer is familiar. Just some motivation up front about trying to estimate values from measurements would have been helpful. I only got to about the 9 minute mark before I lost interest.

Overall, I thought this was a great video. I had the math understanding to follow the content, although I have never encountered this in a class yet (still in high school), so I am probably not the target audience. However, I still found the video very informative and easy to follow without any prior background of the Kalman Filter. I believe the video was well structured, starting the engine example and slowly increasing in detail. The only small note I had was the sheer number of variables was hard to keep track of (I had to go back a few times), but I believe that someone who has seen a Kalman Filter in classes before hand should have a stronger grasp of these variables. But yeah, that was a great video/explanation!

Normally I don’t like to give a score this low to a high-effort 20 minute video, but having watched it I found the video confusing. I feel that both the Motivation and Clarity of this video need work.

Regarding Motivation: Coming into this video I have no idea what the Kalman filter is. It’s my first time hearing that word. So the way you jumped right into the math did not connect with me. I don’t know what problem Kalman filter tries to solve, or even what field of math we’re doing.

In fact, I looked up Kalman filter myself on Wikipedia and learned that it’s a statistical tool for noise reduction of joint variables. I can’t give you credit for that though because I looked it up on my own.

Then on to the math steps. Again, I had no idea what is the problem you were solving. What is the end goal? What are you demonstrating? There were so many equations. Which one is the actual Kalman filter and which ones are the standard statistics stuff? I felt like it was just step after step and I didn’t know where we were going.

I’ve watched enough niche math videos on YouTube to have absolutely zero doubt that even a topic like Kalman filter could be made very interesting. But the video should present the topic in a way that makes sense to me.

hard core but technical info is too dense while lack of intuition. human voice is also a little flat.

The repetition at the start was too short to be useful. Otherwise, a really nice video, if a bit technical.

Cool idea, but the animations concept is not used to its full potential. There is too much deriving like you would do it on paper. Still, when the animations are present they are very helpful and do their job so I would work on balancing the right amount of hardcore derivations with the appropriate visualizations. I liked the animations of the noise propagation in dynamics a lot!