PDA

View Full Version : Hello there!



jrowe47
02-01-2008, 03:29 PM
Hi! I'm a newcomer to robotics in general, although I have put together children's kits before, and worked on a few electronics projects. I saw a video of phoenix and was inspired. She's truly a beautiful machine.
I'm a web designer and programmer, and have a broad range of tinkering experience, mainly on the software side of things. I have already learned quite a bit about robotics in general from perusing the forums, and I hope to learn enough to create a phoenix type bot, myself.

Just a few questions, to start, as my curiousity is boundless... ;)

How fast could Phoenix be made to run?

The servo cables... can they be custom spliced, or are they a specific gauge to handle the power running through? (I ask this because I'd like to do some thin, hollow carbon fiber legs, and if they can be spliced, I can make the legs lighter and skinnier.)

And finally (for now) is there software available that would allow me to design a robot? I can do models and whatnot in Blender, and plan on at least putting together the framework for my dream-bot there. I'll probably use an easy 3D framework like python-ogre and it's physics systems to play with motion and so on.

Thanks for any answers, this is an awesome community!

jrowe47
02-02-2008, 02:40 AM
Another question: Can servo controllers be linked together? Say, if I had a pico-itx minicomputer running, could I connect 2 or more of the SSC 32 servocontrollers to it? I'd love to have 96-128 servos (for a hexapod with articulated arms and a tail, in addition to 4 jointed legs.)

Or are servo controllers a mid-level solution, replacing the need for a mainboard of some sort as the controller?

Again, much thanks for any answers :)

DresnerRobotics
02-02-2008, 06:57 AM
Hi! I'm a newcomer to robotics in general, although I have put together children's kits before, and worked on a few electronics projects. I saw a video of phoenix and was inspired. She's truly a beautiful machine.
I'm a web designer and programmer, and have a broad range of tinkering experience, mainly on the software side of things. I have already learned quite a bit about robotics in general from perusing the forums, and I hope to learn enough to create a phoenix type bot, myself.

Just a few questions, to start, as my curiousity is boundless... ;)

Welcome to the community! We're always glad to have new people! Zenta is the creator of the Phoenix hexapod, and is a member of the community here, so feel free to harass him as much as you like :veryhappy: Just dont tell him I sent you! Also on a side note: I'm not sure if you intended to build your own from scratch, but I do believe Zenta is working with lynxmotion in creating a Phoenix kit.



How fast could Phoenix be made to run?Well, Zenta would probably be the person to best answer this question, but I can tell you that based upon my past experience with walking robots (I've built 2 hexapods, and 3 quadrapods total), they are not usually very fast compared to a wheeled or tracked robot. If I had to make a guesstimate I'd say no faster then 1 foot per second max.



The servo cables... can they be custom spliced, or are they a specific gauge to handle the power running through? (I ask this because I'd like to do some thin, hollow carbon fiber legs, and if they can be spliced, I can make the legs lighter and skinnier.)Not sure exactly what you mean by spliced here... the wires can be cut/extended/seperated, but I wouldn't recommend going with any thinner gauge wires.



And finally (for now) is there software available that would allow me to design a robot? I can do models and whatnot in Blender, and plan on at least putting together the framework for my dream-bot there. I'll probably use an easy 3D framework like python-ogre and it's physics systems to play with motion and so on.I believe Microsoft Robotics Studio might have what you're looking for. I'm not sure what format of 3d file it prefers, but I know you can upload/create virtual 3d models of your robots, and run them through simulated environments/scenarios that use the Ageia physics engine.



Can servo controllers be linked together? Say, if I had a pico-itx minicomputer running, could I connect 2 or more of the SSC 32 servocontrollers to it? I'd love to have 96-128 servos (for a hexapod with articulated arms and a tail, in addition to 4 jointed legs.)

I don't recall seeing anything specifically indicating that the SSC-32 can be linked together. However, I do know that you can have more then one controlled by a single PC, given you have enough serial ports to do so (the pico-itx board only has one, so you'd have to use a usb > serial convertor for more).

While using 100ish servos in a project is ambitious, I will warn you that in reality it is not always feasible. This is for a couple of reasons. First of all you're going to run into power issues, that many servos are going to require TONS of battery power to run, or rely on a tethered power supply. Secondly, you're going to run into weight issues BIG time. That many servos on a bot would push their torque far beyond what they are capable of, even using the highest end servos available (and in that case, you're looking at lots of $$$). For a walking robot, you have to keep in mind that your payload will be limited, so not only will the leg servos be unable to hold up that many other servos, you wouldn't be able to house any payload for batteries either, having far exceeded your payload capacity in servo/frame weight alone.

My advice to you on a walking robot is to start with what you need to get it walking. If you plan on having a high payload, pick up some high end servos and experiment with the max weight capacity to get a good feel for how much you can add.




Or are servo controllers a mid-level solution, replacing the need for a mainboard of some sort as the controller?

You are correct, essentially a servo controller acts as a mid level solution to free up I/O ports on a microcontroller or PC, allowing you to control many servos from a single I/O channel, and also alleviates some of the processing power and code, as you can control for example, 32 servos with a single serial signal.

If you have any further questions by all means ask. Also- if you have any specific questions about the pico-itx boards, hit me up as I've gotten a good amount of experience with them with my recent project.

Take care and thanks for joining us!

jrowe47
02-02-2008, 02:16 PM
Excellent, thank you! Also, Johnny 5.3 just plain kicks ass.

I think you had mentioned wanting to implement a chatbot like Alice in their program... have you encountered Project Cyn, or aimlpad? It's a fusion of openCyc and an Alice engine, allowing chatterbots to apply basic commonsense reasoning. Since it's opensource, I believe it wouldnt be too much work adding robot specific plugins (maybe a sensed object to stored object routine, or something like that.)

Anyway, much thanks for the replies, exactly what I wanted to know. As far as the bot goes, I'm reaching for the stars, so I'm not gonna limit my design until I have a better grasp of what's going on. Then I'll shave things down to a realistic version (financially and physically.)

So far I'm imagining a steampunk looking 6 legged, 2 armed, tailed bot. It would have stereoscopic vision, 3D accelerometers, GPS, and gyros for positioning, and it would convert visually perceived objects into 3D models (if a profile checksum of some sort wasn't met, so that it wouldn't constantly process the same things, but could handle new objects.)
The tail would be fully prehensile, and allow the bot to latch on to it's owner, much the same way a pet monkey would do. The two arms would end in 3 fingered hands, and be able to carry up to 1 liter of Mt Dew, in each hand. It would be able to carry 4 beer bottles, for trips to the bar.
I'd want the speed to be around 2 meters per second, at least in short bursts. I also want it to jump 4 feet vertically from a standstill (creative spring work, maybe.) The tail would have a usb connector in the end for easy interface.

For brains, I'd acquire as much as I possibly could of the open source trove of AI software. OCR, Augmented reality toys, facial recognition, spatial orientation, all the physical stuff... add ALICE, Cyc, walking algorithms, and some sort of custom Python neural net configured against the weight, power draw, and joint setup of the bot, so that it can experiment in it's mind before it does something, etc.

The design would be a cross between a mobile PC and a hexapod bot with a highly interactive system. I'd like to teach it to type, and give it a blog so that it would record things it found interesting.

Anyway, I know that it's impractical... but i figure if I go all out on the design, then pare it down, I'm not gonna miss out on any cool possibilities. In the end, I'm betting I'll probably have a budget around $3000, around the end of summer.

Oh, had another thought... some sort of heat absorbent flexible tubing that would pipe coolant through the system, and a fan at the front. Respiration and circulation, baby, yeah! Hmm, and a turreted coolant expulsion system toward the back...

kdwyer
02-02-2008, 06:05 PM
Oh, hell, just go for the zircon-encrusted plasma chamber! LOL
(And I thought I had ambitious designs!)
Three grand is quite a bot budget, but you'll never implement all the stuff you want without going completely mad.
I think the best approach may be to get a mobile platform up and running, one that has a bit of excess payload capacity and modular mounting options.
With even half the hardware you'd like, you'll have a programming task ahead of you that would challenge MIT, or NASA. Simple 'response to stimuli' is not always easy, and behavior (advanced responses) quickly balloons into spaghetti code. Getting a team to work on the coding can help, but who wants to share the glory?
Seriously, bud, you're aiming real high (commendable!) but that way lies madness. Incremental improvements to a basic chassis is probably the best way to step towards this goal.

Either way, you deserve a round of applause for trying something so far-reaching and difficult.
Lets put our end-effectors together for jrowe47! Hurrah!
(See you at the asylum!)

Zenta
02-03-2008, 02:19 PM
So far I'm imagining a steampunk looking 6 legged, 2 armed, tailed bot. It would have stereoscopic vision, 3D accelerometers, GPS, and gyros for positioning, and it would convert visually perceived objects into 3D models (if a profile checksum of some sort wasn't met, so that it wouldn't constantly process the same things, but could handle new objects.)
The tail would be fully prehensile, and allow the bot to latch on to it's owner, much the same way a pet monkey would do. The two arms would end in 3 fingered hands, and be able to carry up to 1 liter of Mt Dew, in each hand. It would be able to carry 4 beer bottles, for trips to the bar.
I'd want the speed to be around 2 meters per second, at least in short bursts. I also want it to jump 4 feet vertically from a standstill (creative spring work, maybe.) The tail would have a usb connector in the end for easy interface.

Hi!
WOW, a VERY ambitious project you got there! :robotsurprised: Seriously I don't think you will be able to make a hexapod using any RC or robot servos that would be able to run 2 meters per secound and jump 4 feet ;). But I'm certain of that you'll be able to make a really awesome hexapod with a budget of 3000$. I've not measured the walking speed of my Phoenix, the speed depends of the walking gait (tripod gait is the fastest, I'm mostly using ripple) and of course how fast your servos are.

Good luck!

DresnerRobotics
02-03-2008, 02:32 PM
Yar, I commend the high ambitions, but you're talking about a robot that I don't think is possible given the current level of technology. Biggest limitation is that servos simply aren't that fast or strong... yet. A $3000 budget is a high one, but you'd be surprised how quickly it can go. I haven't kept a close tab, but I'd wager my J5 project is now running over 3k and isn't even finished.

Again- I don't mean to try to rain on your parade at all, by all means set your sights high and aim for the stars, just keep in mind that while imagination and ingenuity are limitless, sometimes technology itself can hold us back.

jrowe47
02-03-2008, 11:39 PM
I've been looking around, and I thought a good place to start would be considering different hexapod designs, as far as legs go. Tiger Beetles are the fastest 6 legged insects, and the way their legs work is very interesting. The engineering challenge, I guess, would be to model the legs, their range of motion, and then maximize the power output.

And this is all dream-work until summer, I'm gonna save up for the big one. I think I'm gonna buy a few kits before I go for a big project, just to get familiarized with this stuff :)

I've been looking at those linear actuators, and 2", 110 lbs in one second seems to be quite a bit of force... which would be slightly reduced due to leverage, but multiple actuators would allow greater force to be applied, right?

Imagine 2 quarters attached to the opposite ends of 4 equidistant rods. Attach the assembly to the body of a bot. With one of each opposing pair attached to a linear actuator, so that when a rod moved back and forth, the assembly would tilt accordingly, and by using combinations of the two actuators, you'd get a hemispherical range of motion centered at the connection point. The size of the plate is determined by the range of the actuator's stroke, which could also be modified by changing the angle of attachment (which would also change the base of the hemisphere.)

I think a setup like that would gain the fastest limb motion, but are actuators very precise? Or is it a matter of timing? If it can be as precisely controllable as a servo, then I think I'll go with this type of set-up.

Blue is vertical, yellow is horizontal.

Alex
02-04-2008, 08:53 AM
hey jrowe47, welcome to the community:)


I've been looking at those linear actuators, and 2", 110 lbs in one second seems to be quite a bit of force... which would be slightly reduced due to leverage, but multiple actuators would allow greater force to be applied, right?

Imagine 2 quarters attached to the opposite ends of 4 equidistant rods. Attach [...] the angle of attachment (which would also change the base of the hemisphere.)

I think a setup like that would gain the fastest limb motion, but are actuators very precise? Or is it a matter of timing? If it can be as precisely controllable as a servo, then I think I'll go with this type of set-up.


The actuators are pretty precise, but judging by what you mentioned in this post (http://forums.trossenrobotics.com/showpost.php?p=5949&postcount=4) regarding the speed you want to achieve with this robot, I don't think you want to go with the linear actuators. They're really slow, actually I think they're even slower than most servos. I can test one for you, but I know they're way to slow to do what you want. Maybe you're thinking of a pneumatic/hydraulic system? But, that would require compressors or an engine...

I'm loving your ambition with this project!

Have you looked into the feasibility of using stepper motors? I could be totally wrong here (I'm still a newbie to steppers), so maybe others can help chime in on what I'm getting at. Stepper motors are slick. They're really quick & precise, just a tad more difficult to work with than servos. I'm not too sure on their strength & durability though... We just started carrying this stepper motor (http://www.trossenrobotics.com/sparkfun-stepper-motor.aspx) and driver (http://www.trossenrobotics.com/easydriver-v3-stepper-motor-driver.aspx) from sparkfun which may or may not work for you (again, I'm not too sure, just offering another option to look into), but there are a ton of stepper motors and controllers out there so be to do some research.

DresnerRobotics
02-04-2008, 10:47 AM
I don't think stepper motors would work, all the ones I have ever dealt with do not have very much torque. But then again, I've only played with ones pulled from old plot printers. They were big and hefty, but didn't seem to be very strong.

Honestly, I think the only thing that would even come close (and it still wouldn't come anywhere near the proposed speeds, jumping is a whole other issue) is the RX-64's
(http://www.trossenrobotics.com/robotis-dynamixel-rx-64-robot-servo.aspx)
However, for an 18 servo robot, you're looking at over 5 grand in those servos alone, way over the estimated budget.

jrowe47
02-04-2008, 02:35 PM
I think the jumping would require some geared spring work... something that releases all at once, but takesa litle time to coil back up again.

Also, I ran across a 2" stroke linear actuator rated at 250 lbs. Covering 250 lbs of force in 1 second could translate into a geared widget that has a 45 pounds/in torque (250 * .9)/5in) for a single spur gear.

4 of those in the body would power the hind legs of my dreambot, and regular servos would handle the front and ankle joints. 45 lbs of torque would definitely be enough to move quickly, I would hope.

Plastic gears would definitely not handle that very well, so I'll have to look into learning how to gear things, and probably get some gears made of titanium or some other heavy duty material.(karbonite, maybe?)

Also, I have to learn how to stabilize actuators/servos so that while they move a limb, they can't be overly stressed by one.