Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Open Problems in Robotics

scottlocklin.wordpress.com

Post reply on HN
Print view View original

171–180 of 232 posts

Re: Open Problems in Robotics

#171

Post by nopeNopeNooope »

tgflynn wrote: post #154
Animats wrote: post #66

- Motion planning: already discussed. - Multiaxis singularities: much less of a problem than it used to be. We don't need closed-form solutions any more; we have enough CPU power at the robot to deal with this. You need some additional constraint, like "minimize jerk" when you have too many degrees of freedom. - Simultaneous Location and Mapping. SLAM for short: Getting much better. Things which explore and return a…

I don't really understand why depth estimation using binocular vision is still a problem. I worked on this a bit a number of years ago and I thought I had scene matching working pretty well. The problem is I was trying to make it work without actually having two cameras (ie. on a smartphone where binocular cameras were not available at the time and for the most part still aren't). I was hoping to use the acceleromete…

It's not a problem in the sense that it's unsolved.

It's a problem in the sense that it takes a rack of machinery that consumes a huge amount of power and doesn't fit well in a mobile, battery powered, form factor.

Given that flies do this on uW, obviously there's an approach that is workable, but there's a long path to discovery of what we really need that has to happen.

Kind of like how back in the 90s, everyone KNEW you had to use 64bit floats in neural networks to get anything that worked, and now we have carved all that down to the minimum number of bits to make a problem tractable.

Re: Open Problems in Robotics

#172

Post by MichaelRazum »

mchusma wrote: post #126

I work in robotics. The problem for pretty much everything is one of scale and commercialization. Every one of these issues has been addressed, but almost none have a good and bundled solution. It feels like computers in the 70s. All the pieces are there but there aren't mass produced PCs yet somewhat because there aren't the suppliers to make things easier. This is changing fast. For example, SLAM has many things th…

What do you think about RL learning approaches? Just wondering if that stuff may work in practice. If I'm not wrong Sergey Levine said, that the only problem is sample efficiency. So you would need to simulate the real world, or let your robot brake 1 million time dishes till it learns.

Re: Open Problems in Robotics

#173

Post by Animats »

smilekzs wrote: post #158
R0b0t1 wrote: post #137

Earlier quoted context omitted.

Robotics and embedded just doesn't make that much money. I had to turn down a job offer due to the seriously low counteroffer I was given. Or rather, I was given no counteroffer, just told to pound sand after the market rate I gave apparently insulted them.

If you count autonomous driving vehicles as robotics (I know it's a stretch), the funding (and therefore pay) story is a lot better, although you get the usual startup vs BigTech debate. Source: I work in an ADV company.

The amount of money that's gone into that area without shipping a product is insane.

Re: Open Problems in Robotics

#174

Post by Animats »

alfanick wrote: post #157
tgflynn wrote: post #154

Earlier quoted context omitted.

I don't really understand why depth estimation using binocular vision is still a problem. I worked on this a bit a number of years ago and I thought I had scene matching working pretty well. The problem is I was trying to make it work without actually having two cameras (ie. on a smartphone where binocular cameras were not available at the time and for the most part still aren't). I was hoping to use the acceleromete…

Isn't your "that turned out to be too hard" one of the answers? People tend to think about machine vision in perfect conditions. This is the problem , me thinks. In practice, you will get a lens flare in most important case, or lose a sensor or get some rain. Instead of designing solutions starting from most hardcore edgecases (which are common with humans), robotics researchers tend to provide MVP that works in best…

> Isn't your "that turned out to be too hard" one of the answers?

That's a "we need an expert in filter theory" problem. I had four people crash and burn on that problem when we were building a DARPA Grand Challenge vehicle. Combining GPS, accelerometer, gyro, compass, and odometer data to get position is a hard problem. All those sensors are noisy, but in quite different ways. There are off the shelf solutions now, but there were not in 2004. We could not get below 3 degrees of heading noise, and had trouble keeping the sensor map aligned to the real world.

Re: Open Problems in Robotics

#175

Post by dorgo »

YeGoblynQueenne wrote: post #91
alwahi wrote: post #19

Earlier quoted context omitted.

sorry for sidetracking your answer but what actually does convex mean in the context of optimization. I remember looking at a book called convex optimization. Your statement that > Motion planning is very much non-convex, suggests to me that you are very much talking about the same thing. I understand convexity as in a shape. Why is convex good and concave bad in terms of optimization? I don't want you to dumb down t…

Convex/ non-convex optimisation refers to the shape of the error function we're trying to optimise. In convex optimisation we can assume it's, well, convex: . . ε \ / \ / \ / \ / '._ _.' ^ Global optimum In non-convex optimisation we can't make any assumption about the shape of the error function: ,--. .--. ε / \ / \ ,--. / \ / \ / \ / .' \ / \ / \ / \ / \ / \ / `--' \ / `--' ^ \ / ^ | `--' | | ^ | | | | | | | `-----…

>Generally we prefer to find a _global_ minimum of the error function because then we can expect the resulting approximator to generalise better to data that was not available during training.

Sorry to nitpick, but is this true? We are doing optimization here and a global minimum is just a better solution than a non-global minimum. Is there a connection to generalisation here?

Re: Open Problems in Robotics

#176

Post by m0zg »

b3kart wrote: post #90
m0zg wrote: post #14

Earlier quoted context omitted.

> Without an error bound, ML can’t be in charge of anything that could put human lives at risk Humans don't have "error bounds" either, and you trust them just fine.

Two counterpoints here: 1) Humans can estimate their own uncertainty. Ask a person to show how long a meter is, and they'll give you an estimate. Then ask them to show you the "error bounds", i.e. what they're "quite certain" the meter is longer than and shorter than. You are likely to get sensible bounds. Now, humans aren't amazing at this, but the brain does have capacity for estimating how uncertain it is. 2) No,…

Ask a programmer to estimate how long it'll take them to code something to see if we really have error bars on complex functions. The margin will be so wide as to be completely useless.

We do have error bars on simple measurements already. They're right there in the data sheet for the sensor. What you're asking for are error bars on things several levels distant in the layers of abstraction. Humans suck at that. We only cope the same way machines do: through constant negative feedback.

Re: Open Problems in Robotics

#177

Post by jandrese »

Animats wrote: post #173
smilekzs wrote: post #158

Earlier quoted context omitted.

If you count autonomous driving vehicles as robotics (I know it's a stretch), the funding (and therefore pay) story is a lot better, although you get the usual startup vs BigTech debate. Source: I work in an ADV company.

The amount of money that's gone into that area without shipping a product is insane.

The potential upside for the company that gets it right is enormous. Billions of people are tired of wasting their time driving. Entire industries can be built on the technology if it works well.

That said it's a problem that steers awfully close to needing a full real AI and that's been a showstopper for loads of potential solutions for decades now.

Re: Open Problems in Robotics

#178

Post by dorgo »

supportlocal4h wrote: post #129
contingencies wrote: post #52

Robotics founder here. Popular conceptions of "robots" are unrealistically general. In industry, we do not build robots, we build automation systems. Given the choice, would you prefer an automation system with some environmental assumptions, high speed and perfect repeatability (ie. entire industrial automation world), or no environmental assumptions, crushingly high cost, slow speed and poor reliability (eg. walkin…

It intrigues me that most people do not consider a refrigerator to be a robot. Same for a backhoe or a blender. The distinction between "machine" and "robot" is interesting. A traditional tractor is not a robot, but a driverless tractor is?

Could it be that a refrigerator is a final-state-machine and a robot is higher in the hierarchy?

https://en.wikipedia.org/wiki/Finite-state_machine

https://en.wikipedia.org/wiki/Automata_theory

Re: Open Problems in Robotics

#179

Post by dorgo »

contingencies wrote: post #52

Robotics founder here. Popular conceptions of "robots" are unrealistically general. In industry, we do not build robots, we build automation systems. Given the choice, would you prefer an automation system with some environmental assumptions, high speed and perfect repeatability (ie. entire industrial automation world), or no environmental assumptions, crushingly high cost, slow speed and poor reliability (eg. walkin…

Given the choice, would you prefer an automation system with some environmental assumptions, high speed and perfect repeatability, or less environmental assumptions, slightly higher cost, slower speed and poorer reliability?

Re: Open Problems in Robotics

#180

Post by liability »

cptnapalm wrote: post #135
treeman79 wrote: post #105

Earlier quoted context omitted.

Lost binocular vision for a number of years back. You get around just fine without it. Got it back after some speciality glasses (PRISM). Was a complete shock see depth again. Didn’t seem to help anything getting it back. Mostly just trippy. Chairs were amazing to stare at.

From a much shorter and very different route, once after an 18 hour straight Quake marathon, I looked around the my room and was startled at how everything looked. I was highly impressed by the graphics and depth perception.

This reminds me of the 'Tetris effect' in which people who play Tetris for prolonged periods of time will begin to experience Tetris-like hallucinations when they stop. I've experienced it myself and have trouble explaining it, but it's as though everything you see becomes Tetris-like in some way. You look at your dinner and see ways to rearrange the peas so they 'fit' with the mashed potatoes, or something like that. There's more to it than that though, there's a real sensation of things being moving blocks that must be fitted together. It's bizarre.

But unlike Quake, Tetris isn't providing you with a 3D experience. That makes me wonder how much overlap there is between the two experiences.

Post reply on HN