That looks pretty cool. Mind if I ask what you gunna do with an aluminum skull? It seems too big and heavy to stick onto a humanoid unless you got some 115$ servos You should have gotten a smaller speaker and put it in the back rather than the bottom. Without the hole in the bottom, you cant have wires come out of it. it must come out from the back which doesnt look as cool. Canāt wait to see a final working product
its actually really small, the heaad i mean, and aluminum is extrmely light
The head is fairly light. It does add considerably more weight than say just the PING sensor, or a GP2D12 IR Sensor, but not so much that you need mega servos everywhere.
I agree the design could have been better, I would have liked the wires all come out the back in one area, but then again, try making something from nothing and getting it right the first time
The speaker in the back idea would be difficult to design in a practical way. The bottom mounted speaker allows for a larger diameter speaker to be used. The larger the speakerĆ¢ā¬Ā¦ the better the sound quality. When you get into small speakers, you start to get ābuzzy beeā like sound qualities.
My next design is taking all the bad points of this design and improving them.
WOW
mike dude YOU ROCK!!! cool head
unfortunataly i canāt use it in project espio but
now iām more driven to build it also can you tell
me where you got the software for the concept
design
oh yeah about the cmos cameraās can you tell me the type for
ohā¦letās say a quarter-scale humanoid
I would also like to know that.
Sorry, I have been away from the forums lately.
There are several different types of cmos cameras to choose from. It really boils down to your needs. You might want to check www.supercircuits.com to shop to see the different types. When you mention quarter scale biped, did you mean quarter the size of the lynxmotion biped?
You also mentioned you wanted to know what software that I used to design the head: I used eMachineShop to design it. That can be found here:
Hope this helps.
yes actually iām using hs805bb servos also is there brackets any for this servo and im using all 32 ports plus i need more any ideas
btw thanks for the links they were a big help
Uhā¦
32 of those monsters?!?!
Iād strongly advise avoiding that.
If my meterās current function were still working (I accidently replaced it with a too-high value fuse), I could give you a better guess of how scary an idea that is.
Letās just say that Iām suspicious that these buggers surge at least an amp on startup.
They probably draw around a half amp under normal strain.
What does that mean?
That means youāll need a really heavy battery and really thick wire capable of sourcing 16A or more.
Yikes!
Is the hs805bb really that much of a monster of a servo? I mean the 5955 rates for almost as much torque. I mean, I was hoping that for the added size the servo might be significantly stronger, as well as slightly faster. If you guys know any mega servos that have 500+ oz/in of torque (besides industrial ones which Iāve seen for 1500 oz/in) it would be nice to share for future reference (donāt have any large scale projects planned), but simple searching didnāt show me much.
This beast is larger, which means (usually) a bigger motor.
A bigger motors tends to be stronger but draw more current.
The motors in the 5995/5955 are an exception because theyāre specifically manufactured for an insanely high efficiency.
Which means, you get more power for less current.
Efficiency is almost always reflected in price.
Note the cost differences.
Iām not saying that you shouldnāt use themā¦
Just be aware that youāll need an excellent (and probably heavy, comparably) power system.
Now I know what he ment by quater scale biped. He means quater scale of a full size human! I thought he was talking quater scale of the lynxmotion sized biped.
1/4 scale humanā¦ about a 3yroā¦ good luck getting the robot version to do what you want better than most human 3yro will do.
Thatās an insanely large and expensive project. Isnāt that larger than the Sony Qiro?
Uhā¦
32 of those monsters?!?!Iād strongly advise avoiding that.
If my meterās current function were still working (I accidently replaced it with a too-high value fuse), I could give you a better guess of how scary an idea that is.Letās just say that Iām suspicious that these buggers surge at least an amp on startup.
They probably draw around a half amp under normal strain.What does that mean?
That means youāll need a really heavy battery and really thick wire capable of sourcing 16A or more.
Yikes!
not quite 32 hs805ās but only 4 of them would that be a problem
Thatās an insanely large and expensive project. Isnāt that larger than the Sony Qiro?
yes its larger but less expensive
now a queston is there some vision circut using a basic stamp
and where can i get a microprocessor iāll need 6 of them
Not sure how youāre going to do useful image recognition with a stamp, but you probably want the CMU interface. Here is an example of it doing very basic things like recognizing a bright spot:
acroname.com/robotics/info/examples/cmucam-1/cmucam-1.html
It isnāt for sale there anymore, and the Appmod version from Parallax really limits your design freedom.
Considering I want to use real time image recognition I would be very interested if you could do any of this with a stamp. I want my Scout to be able to do things like avoid obstacles, identify common objects around it (like a computer tower or its power charger) and hopefully even be able to identify if something is a person. Maybe if Iām really lucky I can get it to recognize who it is with reasonable (>50%) accuracy. However, without offloading the calculations to a cluster of computers or worse yet custom DSPs (which is what MITās COG does) I sincerely doubt I will get much further than some advanced optical avoidance, basically turning a camera and pan and tilt into a high resolution ping. Technically I do have access to a computational cluster (though my research advisor might not like the project) but Iād also like to be able to use the bot independently. In my mind for some reason image processing should take at least a 1 Ghz PIII. I think this vaguely comes from the experience of some of my friends doing high-speed molecular interaction photography in an adjacent lab. They eventually needed a cluster of DSPs for their work but are currently going at around 1000 frames per second.
Not sure how youāre going to do useful image recognition with a stamp, but you probably want the CMU interface. Here is an example of it doing very basic things like recognizing a bright spot:
acroname.com/robotics/info/examples/cmucam-1/cmucam-1.htmlIt isnāt for sale there anymore, and the Appmod version from Parallax really limits your design freedom.
Considering I want to use real time image recognition I would be very interested if you could do any of this with a stamp. I want my Scout to be able to do things like avoid obstacles, identify common objects around it (like a computer tower or its power charger) and hopefully even be able to identify if something is a person. Maybe if Iām really lucky I can get it to recognize who it is with reasonable (>50%) accuracy. However, without offloading the calculations to a cluster of computers or worse yet custom DSPs (which is what MITās COG does) I sincerely doubt I will get much further than some advanced optical avoidance, basically turning a camera and pan and tilt into a high resolution ping. Technically I do have access to a computational cluster (though my research advisor might not like the project) but Iād also like to be able to use the bot independently. In my mind for some reason image processing should take at least a 1 Ghz PIII. I think this vaguely comes from the experience of some of my friends doing high-speed molecular interaction photography in an adjacent lab. They eventually needed a cluster of DSPs for their work but are currently going at around 1000 frames per second.
from the way it sounds its impossible i mean it seem simple with the link
you gave me thanks btw but with my limited programming skills guess
i cant do itā¦but maybe i can get help from sn96 and his web page i mean
the guy came up with his rscu board im sure heāll come up with somthing
or have some killer sweet ideasā¦i hope
Mikeāll probably point you towards getting a wireless camera, a video capture card for your PC, and offloading the processing to the RoboRealm software.
Itās probably the best cheap-and-easy way to go (the software is FREE).
Thereās no real programming required.
This, of course bypasses the STAMP entirely, but you could always get another wireless module and make a simple windows program to send back instructions to your STAMP.
In that way, you could have your PC make decisions based upon what it āseesā and then tell the bot where to go from there.
Itās a pretty good choice for something like a biped where weight and space are really an issue.