On PID: the field has been stuck trying to do analytical/optimization-based control for decades, and end-to-end control has shown incredible performance (e.g., SoTA cost of transport in legged locomotion) and robustness (e.g., not falling over when stepping on a pile of leaves) - while being far scalable (in terms of how fast it is to get a new robot up and running). Which is not to say it's perfect but it seems like it's a step in the right direction.

Better yet, companies like Physical Intelligence are doing good with hierarchical ("fast-slow") architecture to address both the intelligence and latency fronts.

I do think robotics would come up with more safety mechanisms (provably safe motion planning etc) just because the risk is a lot more serious than LLMs spitting half-truths

I spent the last few months building a nice frontend (https://oimg.org | https://github.com/yunho-c/OIMG) to latest-and-greatest image codecs¹, because my computer was always full² and images were taking up a big chunk of it.

https://assets.yunhocho.com/images/MacBongPro-Storage.png

I just reached a stopping point for the MVP and published OIMG to Microsoft Store and App Store. And before posting about it, I came to seriously question how many people will actually find it useful... because I don't want to waste their time showing something they won't find useful.

So the question is: how much do images take up in your PC, either in percentage or absolute size?

¹ like jpegli, oxipng, libjxl, etc—that can generally shave 30~50% off of typical PNGs and JPGs either losslessly or with visually imperceptible differences, respectively

² despite offloading a lot of images and files to the cloud regularly. Apple's storage pricing really doesn't help here

Also, if anyone does find OIMG potentially useful, and have any suggestions or feedback, I'd really appreciate it!

Thanks all!

Comments URL: https://news.ycombinator.com/item?id=48961047

Points: 1

# Comments: 1