Ok, so 8 is a number I picked out of the air, however, due to their modular capabilities allowing you to connect them together and get even more processing power, what would you do with x XK-1's?
What would you make? How many would you need/use?
I was thinking about humanoid robotics (I am not hugely experienced in this area, but something I like to think about), would it be cool to have an XMP-64 as the "brain" and then have several XK-1s connected via XLinks throughout the rest of the body to perform localized processing for specific tasks? E.g. I/O etc and if it is important pass it back to the brain to process and combine with other inputs etc etc
Just curious as many of you had some great ideas for the XMP-64 thread.
What would you do with 8 XK-1's?
Well you could have one XK-1 control one limb (Drive the motors and receive any position or toque feedback) I could also take care of sensors to detect objects around it.
Building a humanoid robot is not very easy especially if its full sized.
Building a humanoid robot is not very easy especially if its full sized.
Indeed it is not easy, I was just thinking about how you could utilise the above to aid you in such a situation - as you said you could utilise an XK1 for each limb. Maybe have extra sensors attached too for proximity and touch or something.
Humanoid robots seem to be all the craze at the moment, so naturally my brain went off on a ponder.
It is good to think out loud and brainstorm ideas, what seems complex now, will be potentially simpler in the future.
My background as a computer scientist has made me try and make everything modular from my software development days. Maybe the future will allow us to build a robot by being able to buy an arm from here, a hand from there, mind/logic from somewhere else, all modular, and then they all connect up and just work. Maybe using evolutionary algorithms it figures out how to walk and utilise its components for itself... OK a bit of an ambitious dream, but you never know, things are certainly heading that way.
Humanoid robots seem to be all the craze at the moment, so naturally my brain went off on a ponder.
It is good to think out loud and brainstorm ideas, what seems complex now, will be potentially simpler in the future.
My background as a computer scientist has made me try and make everything modular from my software development days. Maybe the future will allow us to build a robot by being able to buy an arm from here, a hand from there, mind/logic from somewhere else, all modular, and then they all connect up and just work. Maybe using evolutionary algorithms it figures out how to walk and utilise its components for itself... OK a bit of an ambitious dream, but you never know, things are certainly heading that way.
I have plans to integrate a second XK-1 into my ROV project but thats still a month or so down the road... with 8 XK-1s and a boat load of 8x8 RGB LED matrixs you could probably make yourself a decent sized graphics display, this has been similarly discussed in the 3d/raytrace idea... I still say someone should step up and write some wpa/wep cracking code.
Maybe one of those modular robots where each robot is a little cube (etc) and they can join themselves together in a number of shapes to make a larger synergistic robot.
The XK-1 would be perfect for something like that, each bot has a powerful brain and they are ready for high speed linking and distributed processing etc.
The XK-1 would be perfect for something like that, each bot has a powerful brain and they are ready for high speed linking and distributed processing etc.
Well, there would be a lot of fun things that you could do with 8 XK-1's IF you would be able to have the damn things talking to each other over their Xlinks!
I have 4 XK-1's and was never able to get more than 2 talking to each other and that only with some tricks from Martin Young.
The other 'trick' with pull down resistors never worked for me:
http://www.xmoslinkers.org/forum/viewto ... ?f=6&t=397
Gerrit
I have 4 XK-1's and was never able to get more than 2 talking to each other and that only with some tricks from Martin Young.
The other 'trick' with pull down resistors never worked for me:
http://www.xmoslinkers.org/forum/viewto ... ?f=6&t=397
Gerrit
Looks like it works here - they all change and execute new code...
Oops, apologize, got XC-1 and XK-1 mixed up.
XC-1's are the damn things that are reluctant to talk to each other.
XC-1's are the damn things that are reluctant to talk to each other.
Ah ok. I have not tried that.
If you say x XK-1's then that answer would be easy.
If a more bound enviorment of 8 XK-1's I would diadactically
connect 2 xcore as one with 6 5bit xcore I/O leaving a 2-bit
xcore channel for control and global broadcast. Then like it were
easy, the other six cores with 6 5bit xcore I/O and 2bit control,
would link to the diadactic couple, one 5bit link each, leaving 5 5bit
xcore links for the 6. Next connect 4 of the 5 5bit xcore links to nearest
4 xcores leaving 1 5bit xcore for I/O per core. That way you could continue
to expand along 3 Axis bidirectionally, else connect 6 remaining 5bit ports to
there opposites for a more ridgid closed system. now do I connect the 2bits in
daisy or bring them to the front for route and monitor. Yah that should do it, now
what to do with the non event side.
xXK-1 would assume x$$$,$$$,$$$.00 Therfore without limits I'd hire the talent from
this site to creat an online Xbase machine. Visualization, immersion, based on CSP,
something that would make secondlife of WOW seem trivial ; ) Something to exploit
raw data for a more meaning. Sites built with Xcore would be cost efficient and provide
people with features, not so easely or evenly possible. You ask me "What would I do wth it?"
Thats easy to answer, I'd build it, thats my job. I can't answer the real "What" that's totally in
the hand of the developer community, its there job. They'll have fun and create 'What" I could
never dream.
XCore as Dream Machine.
XDM
Shawn
If a more bound enviorment of 8 XK-1's I would diadactically
connect 2 xcore as one with 6 5bit xcore I/O leaving a 2-bit
xcore channel for control and global broadcast. Then like it were
easy, the other six cores with 6 5bit xcore I/O and 2bit control,
would link to the diadactic couple, one 5bit link each, leaving 5 5bit
xcore links for the 6. Next connect 4 of the 5 5bit xcore links to nearest
4 xcores leaving 1 5bit xcore for I/O per core. That way you could continue
to expand along 3 Axis bidirectionally, else connect 6 remaining 5bit ports to
there opposites for a more ridgid closed system. now do I connect the 2bits in
daisy or bring them to the front for route and monitor. Yah that should do it, now
what to do with the non event side.
xXK-1 would assume x$$$,$$$,$$$.00 Therfore without limits I'd hire the talent from
this site to creat an online Xbase machine. Visualization, immersion, based on CSP,
something that would make secondlife of WOW seem trivial ; ) Something to exploit
raw data for a more meaning. Sites built with Xcore would be cost efficient and provide
people with features, not so easely or evenly possible. You ask me "What would I do wth it?"
Thats easy to answer, I'd build it, thats my job. I can't answer the real "What" that's totally in
the hand of the developer community, its there job. They'll have fun and create 'What" I could
never dream.
XCore as Dream Machine.
XDM
Shawn