| Author |
Message
|
| Frenchi |
 Posted: Sat Jul 12, 2008 4:05 pm Post subject: Computers in Robots |
 |
|
Forum Sophomore
Joined: 06 Mar 2008
Posts: 134
Location: Port Saint Lucie, Florida
|
Not sure if this belongs in electronics or computer science... When dealing with robotics, installing an entire PC as a control mechanism seems impractical considering size, heat, and power-consumption parameters, at least in robots that aren't intended to run with AC-power and control wires hanging off of them. For more self-contained builds, what kind of control hardware is common? I'm sure this probably varies by application, but any insight is welcome information =)
Just off the top of my head, I can imagine one would need at least a 'motherboard' with some kind of processor on it, whether it's an Intel/AMD type PC processor or one specifically tailored to controlling servos and the like. Power moderation should be important in extending battery life, so I would reason that PC processors might be a bit overkill for such an implementation, yet while looking at builds of some robots, I see a number of them running specialized Linux distributions. Any knowledge about this? |
|
| Back to top |
|
 |
| CelticMadScientist |
| Posted: Sat Jul 12, 2008 5:43 pm Post subject: |
|
|
Forum Freshman
Joined: 12 Jul 2008
Posts: 19
Location: U.S.A.
|
Hee hee, hello again. My specialization for my M.S/Ph.D is supposed to be control and my stuff revolves around robotics, though I haven't really built and actual robot so much.
One thing to note, control kind of has 2 areas (not officially necessarily, just as I see it). First, you have to make sure things do what you tell them. That seems straightforward, but it's not. It gets really tough sometimes, and involves signal-processing type math, like fourier transforms and stuff (but from your other thread, sounds like your thing). For example, you can't just attach a DC motor to a robotic arm segment and apply a burst of voltage to send it to a position. You have to continually conteract gravity, anything it might be pulling, or if something touches it - in other words make it "robust." So, quite simply, you might in this case add a variable resistor (potentiometer) to the motor axis and feedback a signal that gets amplified and subtracted from your input angle voltage. That way, if gravity starts to pull it down, it applies more voltage to bring it back to the desired angle. This package is called a servo motor, which you can buy.
The more fun side (from my view) and more what you're talking about it a kind of more software side of things. As in, "ok, I have the above device (servo motor), I just want to tell the robot to move here, at this speed, at this time, then do this, etc." And you're right, depending on the system requirements: portability, power consumption, weight - you're probably not always going to use a full fledged Dell Pentium machine. You might use an "embedded device." Common things are programmable microntrollers - a package of microprocessor plus some extra hardware to interface with the outside world. You might look in PIC. They're common and cheap. I have a kit here that can program flash memory microcontrollers, so you can really debug them. They have input ports to receive feedback, buttons presses, whatever & outputs that could be used to drive LEDs or motors (with added power circuitry). A lot of the time for small stuff you program them with really low level code (assembler), but more and more you can program in C and have it compiled like on a PC.
A step up from that is hobbyist boards like I used in my freshman honors program. We built autonomous robots to traverse a course of obstacles and complete goals. These things have some of the extra circuitry (like motor ports) already made.
There's other various things, like VLSI stuff and such, but not quite sure what you're looking for.
_________________
Celtic Mad Scientist
Celtic Mad Scientist's MetaCafe Channel - Science & Fun How-to Videos
celticmadscientist.com |
|
| Back to top |
|
|
| Frenchi |
| Posted: Sat Jul 12, 2008 6:07 pm Post subject: |
|
|
Forum Sophomore
Joined: 06 Mar 2008
Posts: 134
Location: Port Saint Lucie, Florida
|
| CelticMadScientist wrote: |
| There's other various things, like VLSI stuff and such, but not quite sure what you're looking for. |
I'm not exactly sure either, I'm just looking for general field knowledge to sponge up. I've been steering in the direction of robotics more or less since I got out of high school, working my way up the knowledge ladder. The field has always struck me as something you don't just jump into without having more fundamental building blocks underneath you, though my focus thus far has definitely been more on the software side than the hardware. Recently I've been studying language constructs, building chat bots and finding ways to extend the things they can do. I'm about to move my most robust one onto its own server, as soon as I can bear parting with the money =)
That aside, while I've been working with computer hardware for many years, I have no idea where to start learning about robot hardware (or robot software, for that matter). Everything I've learned so far has been self-taught, so finding someone more knowledgeable than myself to point me in the right direction is my current focus. |
|
| Back to top |
|
|
| CelticMadScientist |
| Posted: Sun Jul 13, 2008 11:57 am Post subject: |
|
|
Forum Freshman
Joined: 12 Jul 2008
Posts: 19
Location: U.S.A.
|
Well, the more software side of robotics class I took dealt with kinematics, dynmics, motion planning, vision, and a little control. We focused on how to move a 5 link robot arm where it needed to go, we ran simulations of it in a 3D program. There are issues to consider like, with 5 links, you might have more than one configuration of angles to get the end effector where it needs to be. Perhaps forward and inverse kinematics would be a starting point if you wanted to dive in a fair bit.
There's also kits of various levels. You can even get a 3 link robot arm for a moderate amount. Some kits have open endedness, so you can fiddle with the program and add on new sensors.
_________________
Celtic Mad Scientist
Celtic Mad Scientist's MetaCafe Channel - Science & Fun How-to Videos
celticmadscientist.com |
|
| Back to top |
|
|
| Frenchi |
| Posted: Sun Jul 13, 2008 7:27 pm Post subject: |
|
|
Forum Sophomore
Joined: 06 Mar 2008
Posts: 134
Location: Port Saint Lucie, Florida
|
I'm sure I'll need to submit to taking classes at some point, but my work schedule and bank account aren't permitting currently, so I'm stuck at hobby status. I don't mind picking up a book to continue my education, but finding one in an area of interest that doesn't go over my head has proved troublesome =)
| CelticMadScientist wrote: |
| Perhaps forward and inverse kinematics would be a starting point if you wanted to dive in a fair bit. |
Thanks, I'll tack this to my wall of things to research 
A kit wouldn't be a bad idea, up until now I've been disassembling robots from Toys'R'Us to see how they tick. |
|
| Back to top |
|
|
| CelticMadScientist |
| Posted: Mon Jul 14, 2008 11:55 pm Post subject: |
|
|
Forum Freshman
Joined: 12 Jul 2008
Posts: 19
Location: U.S.A.
|
|
| Back to top |
|
|
|
|
|
