Have a look at the Servo 32 object and the Memsic dual-axis accelerometer object in the Propeller Object Exchange. The Propeller Tool comes with some A/D conversion routines to use with the built-in microphone on the Demo Board. The same circuit can be used on the Proto Board if you're very careful about lead lengths ... Read the application note on the Propeller's counters for more information on this (from the Propeller download page).
You have a lot of work and learning ahead of you. You'll have to learn enough to extend the dual-axis accelerometer code to handle the triple-axis version. You'll need to use the existing A/D code and microphone examples to write your own direction of sound detectors. You'll need to develop the code to do the mid-level management of the movement of the robot.
You will probably want some kind of wireless link. Consider xBee or Bluetooth. Check www.selmaware.com for xBee and Parallax's Embedded Blue 500 for Bluetooth.
You will probably want some kind of mass storage. Check out Rokicki's SD card support routines in the Propeller Object Exchange. Also look at FTDI's Vinculum devices (www.vinculum.com) and GHI Electronic's uALFAT and USBwiz devices.
The AI that I am hoping to create is the raptor being able to cross a situation (say laying on my lap) to an action (say petting the robot). Any ideas for that?
What type of sensor could detect if the robot is being lifted or if it is on its side.
What type of sensor would allow the raptor to know in what position all of its continous rotation servos are?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
1) As many as you can handle. They need to communicate with each other. There are networking techniques that can work with one I/O pin that can handle a hundred or more Propellers. You only have to do the programming for it. The techniques have been used on slower computers with less memory than the Propeller.
2) Nope
3) I don't know about being lifted. The accelerometer could detect position.
4) You would need an encoder on each servo. These produce pulses as the servo rotates and the Propeller can count the pulses to determine the position of the servo. There's a nice description on the Wikipedia (en.wikipedia.org/wiki/Rotary_encoder)
1) Mike is right... as usual. And you HAVE asked the $64,000 question;>)
2) There is no such thing as artificial intelligence... either your raptor will be intelligent or not. Right now, I think you are going to want a computer to do the thinking... and the propeller to do the sensing and actions. Communications is going to be the fun part. And that is where I would start if I were you... you have all kinds of choices ... infra-red, with multiple objects posted, R/C, with multiple objects posted, xbee, etc. etc. You just have to pick the one that is easiest for you.
3) You are probably going to want scads of contact switches and pressure sensors... combined with multiple 3d accelerometers, BUT you COULD use a few ultrasonic rangers or even a camera or two with some image processing done on your computer. You could even use a couple of microphones...and some little speakers to measure sound attentuation changes, that's how they caught Lee Harvey Oswald... isn't it?
4) Mike is right again.... but you don't have to use servos... why not use hydraulics? That's how Disney does it.... sometimes. Then you only need one motor and everything becomes a linear measurements, which you can sense in all kinds of ways.
Please post ALL of your thoughts and eventually your solutions.
Thanks,
Rich
PS: If you want your raptor to fly... Graham Stabler has probably done it already.
On Hobby servos you can also tap off of the potentiometer inside them to get a position signal. The potentiometer is an integral part of the pulse feed-back loop so it could be hard to find the right time to read the potentiometer's voltage. Hi-Tec also sells a digital servo that'll allow it's position to be read out. Or just use a bunch of AX-12+ servos from http://www.crustcrawler.com/
Lawson has an interesting suggestion about the newer digital servos. They don't work like standard R/C servos and use asynchronous serial signalling (as far as I can figure) around 1MBaud. The FullDuplexSerial.spin routines can't handle that, but other half-duplex (one way at a time) routines could be written in assembly to talk to them. Read some of the sales info and datasheets on the AX-12 servos. They have some great sounding capabilities.
There is already a propeller based raptor project under way. Check out www.bioloid.info/tiki/tiki-index.php?page=MicroRaptor. It is AX-12 servo based and will have a six-axis IMU www.bioloid.info/tiki/tiki-index.php?page=6-Axis+Bus+IMU. I am also working on a propeller based controller for the Bioloid AX-12 servos. The guy working on the raptor modified the full duplex serial object and got it working. I think I am going to modify the half duplex object used with the HYDRA for mine. The AX-12 is ideal for propeller based robotics. Each servo provides force and position feedback which you will be needing.
You'll need to build a 'spine' with joints that can swivel any which way.
Here's a way to build a simple prototype:
Get a bunch of drinking-straws which can be bent at one end.
Take a few cardboard squares, 2" x 2", poke a hole in the senter and stick a plastic straw through each, making certain that the bendy area is approx .5" - 1" from the cardboard.
Trim the straws so that you have about .5" ends beyond the bendy area.
Cut the other end about .5 - 1" on the other side of the cardboard(to match the area between cardboard and bend).
now, join the straws together until you have a chain of cardboard squares parallell to each other.
Now, mechanizing it...
you can make holes at the top, bottom and sides of all the squares. Then thread rubber bands thorugh the bottom and one of the sides.
The 'spine' will now curl up somewhat...
(If not, just tighten the bands a little)
Add a servo to one end of the 'spine' and a thread from it, through an emty hole on the first square and tie off to the second square.
As soon as the servo is moved one way, that section is straightened, or even bent opposite of what the rubber band wants it to. Moving the servo the other way will make it curl up again.
Add more threads, but these must go through the CENTER of the spine until it gets to the second square, where it is threaded through the hole the first thread was tied off to.
The second thread is then tied off to the third square.
This thread can also be connected to the first servo, or it can be connected to a second servo if you want to control individual joints.
Keep adding threads and servos until you have the right articulation.
Alternate possibilities:
1. Attach all threads to a rod held parallell to the rearmost square, with the first thread on one end and the last at the other end. Then use one servo to move the rod closer to or further away from the spine assembly to flex/unflex it, and another to angle the rod to change how much each joint is flexed.
2. Use an 'X' instead of a rod, and get both X and Y flexing with one servo moving it closer or further away from the spine, and two to angle it for adjusting the flexing.
3. Vary the distance between the threads on the rod/X to account for the joints closer to the body normally having less flexibility because of thicker necks.
4. You may want to mount threads that go from the top of one square to the side of the next, which when pulled will twist the spine-assembly. (Will not work on a model with drinking straws as they don't bend like that)
This may need to have a second thread on the other side which is slacked up instead of rubber bands, though.
It may be possible to use small DC motors instead of servos...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Don't visit my new website...
Bipedguy
Posts: 40
Comments
Hardware: Different voltage and pinouts.
Of course I don't know anything about your raptor yet.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I stand on the shoulders of giants
-The raptor will be a comepletely autonomous bipedal robot which is now based on the Propeller P8X32A-Q44 (QFP) Chip.
-probably 32+ inches long
-AI(Hopefully)
-Mics to detect direction of sound
-3-axis accelerometer
and much more!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
Detective Spooner
iRobot
You have a lot of work and learning ahead of you. You'll have to learn enough to extend the dual-axis accelerometer code to handle the triple-axis version. You'll need to use the existing A/D code and microphone examples to write your own direction of sound detectors. You'll need to develop the code to do the mid-level management of the movement of the robot.
You will probably want some kind of wireless link. Consider xBee or Bluetooth. Check www.selmaware.com for xBee and Parallax's Embedded Blue 500 for Bluetooth.
You will probably want some kind of mass storage. Check out Rokicki's SD card support routines in the Propeller Object Exchange. Also look at FTDI's Vinculum devices (www.vinculum.com) and GHI Electronic's uALFAT and USBwiz devices.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
Detective Spooner
iRobot
How many propellers can work together?
The AI that I am hoping to create is the raptor being able to cross a situation (say laying on my lap) to an action (say petting the robot). Any ideas for that?
What type of sensor could detect if the robot is being lifted or if it is on its side.
What type of sensor would allow the raptor to know in what position all of its continous rotation servos are?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
Detective Spooner
iRobot
2) Nope
3) I don't know about being lifted. The accelerometer could detect position.
4) You would need an encoder on each servo. These produce pulses as the servo rotates and the Propeller can count the pulses to determine the position of the servo. There's a nice description on the Wikipedia (en.wikipedia.org/wiki/Rotary_encoder)
Re: your questions.
1) Mike is right... as usual. And you HAVE asked the $64,000 question;>)
2) There is no such thing as artificial intelligence... either your raptor will be intelligent or not. Right now, I think you are going to want a computer to do the thinking... and the propeller to do the sensing and actions. Communications is going to be the fun part. And that is where I would start if I were you... you have all kinds of choices ... infra-red, with multiple objects posted, R/C, with multiple objects posted, xbee, etc. etc. You just have to pick the one that is easiest for you.
3) You are probably going to want scads of contact switches and pressure sensors... combined with multiple 3d accelerometers, BUT you COULD use a few ultrasonic rangers or even a camera or two with some image processing done on your computer. You could even use a couple of microphones...and some little speakers to measure sound attentuation changes, that's how they caught Lee Harvey Oswald... isn't it?
4) Mike is right again.... but you don't have to use servos... why not use hydraulics? That's how Disney does it.... sometimes. Then you only need one motor and everything becomes a linear measurements, which you can sense in all kinds of ways.
Please post ALL of your thoughts and eventually your solutions.
Thanks,
Rich
PS: If you want your raptor to fly... Graham Stabler has probably done it already.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
Detective Spooner
iRobot
-unmodified servos for legs and most parts of the body.
-XBee Wireless link
-SD support cards
How would I make a neck that could bend from side to side or up and down?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
Detective Spooner
iRobot
There is already a propeller based raptor project under way. Check out www.bioloid.info/tiki/tiki-index.php?page=MicroRaptor. It is AX-12 servo based and will have a six-axis IMU www.bioloid.info/tiki/tiki-index.php?page=6-Axis+Bus+IMU. I am also working on a propeller based controller for the Bioloid AX-12 servos. The guy working on the raptor modified the full duplex serial object and got it working. I think I am going to modify the half duplex object used with the HYDRA for mine. The AX-12 is ideal for propeller based robotics. Each servo provides force and position feedback which you will be needing.
If you are committed on using standard servos check out the PRC board www.wulfden.org/PRC/index.shtml
Hope this helps,
Len
You'll need to build a 'spine' with joints that can swivel any which way.
Here's a way to build a simple prototype:
Get a bunch of drinking-straws which can be bent at one end.
Take a few cardboard squares, 2" x 2", poke a hole in the senter and stick a plastic straw through each, making certain that the bendy area is approx .5" - 1" from the cardboard.
Trim the straws so that you have about .5" ends beyond the bendy area.
Cut the other end about .5 - 1" on the other side of the cardboard(to match the area between cardboard and bend).
now, join the straws together until you have a chain of cardboard squares parallell to each other.
Now, mechanizing it...
you can make holes at the top, bottom and sides of all the squares. Then thread rubber bands thorugh the bottom and one of the sides.
The 'spine' will now curl up somewhat...
(If not, just tighten the bands a little)
Add a servo to one end of the 'spine' and a thread from it, through an emty hole on the first square and tie off to the second square.
As soon as the servo is moved one way, that section is straightened, or even bent opposite of what the rubber band wants it to. Moving the servo the other way will make it curl up again.
Add more threads, but these must go through the CENTER of the spine until it gets to the second square, where it is threaded through the hole the first thread was tied off to.
The second thread is then tied off to the third square.
This thread can also be connected to the first servo, or it can be connected to a second servo if you want to control individual joints.
Keep adding threads and servos until you have the right articulation.
Alternate possibilities:
1. Attach all threads to a rod held parallell to the rearmost square, with the first thread on one end and the last at the other end. Then use one servo to move the rod closer to or further away from the spine assembly to flex/unflex it, and another to angle the rod to change how much each joint is flexed.
2. Use an 'X' instead of a rod, and get both X and Y flexing with one servo moving it closer or further away from the spine, and two to angle it for adjusting the flexing.
3. Vary the distance between the threads on the rod/X to account for the joints closer to the body normally having less flexibility because of thicker necks.
4. You may want to mount threads that go from the top of one square to the side of the next, which when pulled will twist the spine-assembly. (Will not work on a model with drinking straws as they don't bend like that)
This may need to have a second thread on the other side which is slacked up instead of rubber bands, though.
It may be possible to use small DC motors instead of servos...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Don't visit my new website...
Thanks everyone for your great replies.
To the guy who suggested the other raptor, i know (from experience) that a robot based on the AX-12 and uses PCBs goes up in price pretty fast.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
Detective Spooner
iRobot
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
Detective Spooner
iRobot
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Robots are just lights...and clockwork."
Detective Spooner
iRobot