Overclockers Australia Forums

OCAU News - Wiki - QuickLinks - Pix - Sponsors  

Go Back   Overclockers Australia Forums > Other Topics > Science

Notices

Reply
 
Thread Tools
Old 5th June 2011, 10:51 PM   #61
Whisper
Member
 
Whisper's Avatar
 
Join Date: Jun 2001
Location: Sydney
Posts: 8,297
Cool

Quote:
Originally Posted by Resident View Post
They are caused by two things

First, in those videos it is being run from a netbook (all we had available at the time of filming) so it takes longer to process the algorithm.
It also is using a microsoft lifecam HD5000, which we have discovered is quite laggy. This means we have to stop the cube for a split second, wait, then take the photo. We used to run it from a logitech camera that was a lot faster and when we go for the official record we will have put a better camera on it.

Really, it should run at full speed the entire time and you should never see any pauses.
I thought there was plenty room for improvement on the speed of solution if those pauses were removed. Cool
__________________
Quote:
There's something about philosophy that is at once humanising & utterly human. It's the court of last recourse. When the oracles have failed us, when the saints have grown silent & when god has chosen not to reveal himself. Then we stand back in the dark dark shadows of confusion & fear & ask, "What sort of being am I?" "What kind of life is right for me?" "& how should I govern, or be governed?"
Whisper is offline   Reply With Quote
Old 7th June 2011, 8:20 PM   #62
Resident
Member
 
Join Date: Jul 2002
Location: Melbourne
Posts: 123
Default

Quote:
Originally Posted by Cheep_Thrillz View Post
This is nifty!
Did it drop the cube much when you were still developing/debugging it?
Did the cube ever get thrown very far?
I think this answers your questions...
Resident is offline   Reply With Quote
Old 7th June 2011, 9:03 PM   #63
Whisper
Member
 
Whisper's Avatar
 
Join Date: Jun 2001
Location: Sydney
Posts: 8,297
Cool

Quote:
Originally Posted by Resident View Post
Hey, that video shows a 6.0 solution time!!
__________________
Quote:
There's something about philosophy that is at once humanising & utterly human. It's the court of last recourse. When the oracles have failed us, when the saints have grown silent & when god has chosen not to reveal himself. Then we stand back in the dark dark shadows of confusion & fear & ask, "What sort of being am I?" "What kind of life is right for me?" "& how should I govern, or be governed?"
Whisper is offline   Reply With Quote
Old 7th June 2011, 9:08 PM   #64
Resident
Member
 
Join Date: Jul 2002
Location: Melbourne
Posts: 123
Default

Quote:
Originally Posted by Whisper View Post
Hey, that video shows a 6.0 solution time!!
That was early on before we got the vision system going, so no scanning time, just passed a known solution.
Resident is offline   Reply With Quote
Old 7th June 2011, 9:13 PM   #65
dakiller Thread Starter
(Oscillating & Impeding)
 
dakiller's Avatar
 
Join Date: Jun 2001
Location: Gippsland
Posts: 7,334
Default

Quote:
Originally Posted by Resident View Post
I just saw that today, the best part is at the 30 second mark

Quote:
Originally Posted by Whisper View Post
Hey, that video shows a 6.0 solution time!!
Kind of cheating, as the algorithm spits out 22-25 moves, we pre-programmed it with a list of moves and scrambled the cube with these moves so that it would solve back out. It was done so that we would always have the exact same list of moves to test the system with every time we ran it. If we're allowed to look at the cube before solving it and pre-program it in, we are faster than that record holding kid

Last edited by dakiller; 7th June 2011 at 9:16 PM.
dakiller is offline   Reply With Quote
Old 7th June 2011, 9:29 PM   #66
Whisper
Member
 
Whisper's Avatar
 
Join Date: Jun 2001
Location: Sydney
Posts: 8,297
Question

Quote:
Originally Posted by dakiller View Post
Kind of cheating, as the algorithm spits out 22-25 moves, we pre-programmed it with a list of moves and scrambled the cube with these moves so that it would solve back out. It was done so that we would always have the exact same list of moves to test the system with every time we ran it. If we're allowed to look at the cube before solving it and pre-program it in, we are faster than that record holding kid
Are you saying the algorithm can calculate the solution for the entire puzzle from looking at 1 face, or the entire cube?

Then is it just a matter of programming the movements of the robot claws?
__________________
Quote:
There's something about philosophy that is at once humanising & utterly human. It's the court of last recourse. When the oracles have failed us, when the saints have grown silent & when god has chosen not to reveal himself. Then we stand back in the dark dark shadows of confusion & fear & ask, "What sort of being am I?" "What kind of life is right for me?" "& how should I govern, or be governed?"
Whisper is offline   Reply With Quote
Old 7th June 2011, 9:38 PM   #67
dakiller Thread Starter
(Oscillating & Impeding)
 
dakiller's Avatar
 
Join Date: Jun 2001
Location: Gippsland
Posts: 7,334
Default

Quote:
Originally Posted by Whisper View Post
Are you saying the algorithm can calculate the solution for the entire puzzle from looking at 1 face, or the entire cube?

Then is it just a matter of programming the movements of the robot claws?
It has to look at all 6 sides, and once it does then the algorithm that solves it spits out a list of moves in the form of the faces that it needs to turn F,B,L,R,U,D in either clockwise or anticlockwise direction. That is then passed to the embedded system and nothing else needs to be done from the PC end once that is done. As it is only a serial port for communication, you just need to connect to it via a terminal app and send the relevant characters, in a string if you wish, to control the whole thing
dakiller is offline   Reply With Quote
Old 7th June 2011, 9:53 PM   #68
Whisper
Member
 
Whisper's Avatar
 
Join Date: Jun 2001
Location: Sydney
Posts: 8,297
Question

Thanks for that

So solution time is partially a function of processing power.

What's the point where you start hitting serious diminishing returns?

Obviously the netbook doesn't cut it if you were serious about getting the fastest solution, but some how I don't think a super computer is going to help shave off massive amounts of time compared to a decent desktop computer, but I guess that is my question.
__________________
Quote:
There's something about philosophy that is at once humanising & utterly human. It's the court of last recourse. When the oracles have failed us, when the saints have grown silent & when god has chosen not to reveal himself. Then we stand back in the dark dark shadows of confusion & fear & ask, "What sort of being am I?" "What kind of life is right for me?" "& how should I govern, or be governed?"
Whisper is offline   Reply With Quote
Old 7th June 2011, 10:24 PM   #69
dakiller Thread Starter
(Oscillating & Impeding)
 
dakiller's Avatar
 
Join Date: Jun 2001
Location: Gippsland
Posts: 7,334
Default

Quote:
Originally Posted by Whisper View Post
Thanks for that

So solution time is partially a function of processing power.

What's the point where you start hitting serious diminishing returns?

Obviously the netbook doesn't cut it if you were serious about getting the fastest solution, but some how I don't think a super computer is going to help shave off massive amounts of time compared to a decent desktop computer, but I guess that is my question.
We ran it off just a basic desktop PC at uni, they don't have very powerful cpu's in them at all, they are a few years old the desktops and are your average specs. With this, the time to generate a solution is not at all noticeable, as you can see in the first run on the latest video linked above. This algorithm is generating ~22-25 face turn moves to solve, yet it has been shown that any Rubik's cube can be solved in 20 moves or less, just that the time to find these solutions takes the average desktop a few minutes to find. It is just too much effort to shave off those 2 or so moves, when it doesn't take much time for the robot to actually perform the extra moves.

The time to turn the grippers is as fast as 80ms, but we have variable control over this and I think that it has been slowed down slightly from that to be more reliable. The grippers open and close extremely quickly, I think it is in the order of 10ms for that . I remember we did a test that we repetitively opened and closed them about 100 times and it was about a second to do all those moves. There are sensors that tell when it is fully opened and fully closed, so that we aren't just putting delays in, we are actually waiting for it to fully close before the next move is done here.

One spot that we have room for improvement is that while the algorithm counts face turn moves, there is a complex lookup table that converts these face moves into robot gripper moves, and depending on what state the robot is in, it can be up to 7 different actions needed to be performed to make the next face turn move. So what we can do is have the solving algorithm spit out a list of different possible solutions and work out which one has the lowest total gripper move time cost and run with that one.


In looking through the project folder, I came across the step response graph for the motors, which i thought might be of some interest


Click to view full-sized image!
Hosted by UGBox Image Store
X axis is time in seconds and Y is the encoder counts, with 2000 is 90 degrees
dakiller is offline   Reply With Quote
Old 8th June 2011, 11:33 PM   #70
Whisper
Member
 
Whisper's Avatar
 
Join Date: Jun 2001
Location: Sydney
Posts: 8,297
Question



At the 23 second mark in this video there is a noticeable pause.

That is the type of thing I was referring to.

I think I saw others as well, but cbf tracking them down, as I'm not trying to tear your efforts down, just wondering what causes those pauses?
__________________
Quote:
There's something about philosophy that is at once humanising & utterly human. It's the court of last recourse. When the oracles have failed us, when the saints have grown silent & when god has chosen not to reveal himself. Then we stand back in the dark dark shadows of confusion & fear & ask, "What sort of being am I?" "What kind of life is right for me?" "& how should I govern, or be governed?"
Whisper is offline   Reply With Quote
Old 11th June 2011, 8:27 AM   #71
Resident
Member
 
Join Date: Jul 2002
Location: Melbourne
Posts: 123
Default

Quote:
Originally Posted by Whisper View Post
At the 23 second mark in this video there is a noticeable pause.
Looking at that video you can see that the camera side gripper from vertical when that pause occurred, which means that it was caused by the rubber on the gripper tips sticking to the cube for a split second and now allowing the gripper to open. We used to have trouble with that (when that video was taken) as we bought the most grippy rubber we could find, but it ended up acting almost like a suction cup to the surface and sometimes stuck.

We changed to less grippy rubber and it worked a lot better later on.

Problems like that were what made this project hard. There were so many things that you don't consider until it happens and trying to find away around them so they didn't derail the project was often time consuming.
Resident is offline   Reply With Quote
Old 12th June 2011, 1:15 AM   #72
dmr
Member
 
Join Date: Jun 2010
Posts: 147
Default

nice work, good to see we got some local smarties on OCAU.
dmr is offline   Reply With Quote
Old 24th June 2011, 9:24 PM   #73
RatcHeT-x
Member
 
RatcHeT-x's Avatar
 
Join Date: Jun 2011
Location: Brisbane, AU
Posts: 347
Default

Just posting to give you a massive congrats on the record. nice work man!
__________________
OCAU // Drummer Club # [35]
RatcHeT-x is offline   Reply With Quote
Old 1st July 2011, 9:43 PM   #74
Archades
Member
 
Archades's Avatar
 
Join Date: Oct 2005
Location: Jeebus Land
Posts: 1,962
Default

Mechatronics looks damn awesome n fun apart from the mountains of math.

What's the next uni project?
__________________
Flickr | RedBubble
Archades is offline   Reply With Quote
Old 12th November 2016, 3:24 PM   #75
Lasmi
Member
 
Join Date: Jan 2006
Location: Local
Posts: 4,716
Default Holy Thread Revival Batman!!

A new world record has been set for solving a Rubik's cube. In just five years we've gone from 10 seconds to 0.637 seconds.

Robot breaks 'world record' for solving a Rubik's Cube in a fraction of a second

Quote:
A robot has broken the world record for solving a Rubik's Cube in the fastest time ever, according to its creators.

The machine, known as "Sub1 Reloaded" took just 637 thousandths of a second to analyse the toy and make 21 moves, so that each of the cube's sides showed a single colour.

That beats a previous record of 887 thousandths of a second , which was achieved by an earlier version of the same machine using a different processor.

The record attempt was staged at the Electronica trade fair in Munich.
http://www.mirror.co.uk/tech/robot-b...olving-9228939

OP's team's achievement is still incredible, especially just using "two hands", but this is a great example of how far robotics and automation has come in just five years. A different processor alone dropped 250ms off the time. The human record is 4.6 seconds though that's just doing the moves as they get to analyse it first.

This is why there needs to be automated vehicle racing. The advances in F1 don't really filter down to consumer level but with automated vehicles many of the advances will.
Lasmi is offline   Reply With Quote
Reply

Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time now is 7:09 PM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2017, vBulletin Solutions, Inc.
OCAU is not responsible for the content of individual messages posted by others.
Other content copyright Overclockers Australia.
OCAU is hosted by Micron21!