We're not done with point clouds
claytonwramsey.comPosts like these really highlight how much there to think about in robotics. Below the surface of what immediately comes to mind there is just an abundance of topics to address to get an actually well working robot.
Yep. And when you get to an actual robot, you find it’s a distributed system with all the attendant challenges, so you build yourself up for that, and then your time, attention, and maybe your soul are crushed by connectors, cables and batteries.
So naturally the funnest work is multi-robot systems…
Especially your soul, I bet.
well, and your fingers.
In my opinion, it's not just robotics, there are so many processes involved in manufacturing a single high-performance bolt used in an F1 racing car.
Life of Bolt(2016)
https://www.youtube.com/watch?v=iptAkpqjtMQ
Most high quality systems performance relies heavily on high-quality subsystems to achieve optimal performance.
The title implies that there are collision check systems that don’t use point clouds (or that robotics has moved onto something else)? Is that true or did the author just mean personally?
I think they meant it in the sense of coming back to the topic yet again after talking about it fairly exhaustively in an earlier blog post.
Why don’t they marry the visual categorisation of objects with a sparse predefined model to simplify?
Point clouds seem like a noisy layer to use and better as verification.
This reminds me of writing physics in game engines.
When I worked in robotics, it always pissed me off that our recruiters ignored game devs as candidates. There's such a huge overlap in the two fields in terms of the math underpinning the systems
The number of shallow thinkers that believe making video games is playing games all day long and is not incredibly complex software without the people, without the formal educations, and never enough budget to do what one is asked in any reasonable amount of time. And we'd pull it off.
I left games, of course, and after various shenanigans ended up writing facial recognition software. Surprised everyone and became the project lead, nobody was even close in the technical real time simulation requirements space that FR requires.
From what I've seen from the outside looking in, I feel like game development has it's own variation of Wirth's Law where level designers push game engines to their limit faster than the people making the engines can optimize them.