Loading ...
Sorry, an error occurred while loading the content.

Re: SR04

Expand Messages
  • dpa_io
    Hey David, I don t disagree. But PID routines were implemented on my LegoBot and later on SR04 and the other robots not out of curiousity or caprice, but to
    Message 1 of 69 , Feb 4 10:17 PM
    • 0 Attachment
      Hey David,

      I don't disagree.

      But PID routines were implemented on my LegoBot and later on SR04 and the other robots not out of curiousity or caprice, but to solve the very real-world problems outlined below: inconsistent behavoirs as the battery drains, differing behaviors on different surfaces, inabilty to hold a line when crossing different surfaces, and so forth. These are not abstract concepts but real world problems that were encountered in the course of developing the robots. I suspect that you will encounter them too, if you build enough robots and try to get them to behave.

      best,
      dpa


      --- In SeattleRobotics@yahoogroups.com, "David Buckley" <david@...> wrote:
      >
      > Hi David
      > If you understand PIDs and are good at implementing them then of course they can help a lot.
      > My point was PID and shaft encoders are not going to help much in a real world robot of the type most people build or would now like to build eg an SR04 ;). Micromice are not real world robots existing for the very specialized world of the maze. Of course if your error signal comes from other than wheel encoders then that is a different matter.
      > However for those who are not good at implementing PID loops then it shouldn't be seen as a major stumbling block. Hobby robots can work really well without any orthodox PID control.
      > Going back to that web page http://www.speakeasy.org/~tomdean/SRS-Workshop.html
      > It would appear that even with PID control the SRS was almost uncontrollable and enough to put anybody off building a robot. Sure it went in nice straight line but not in the right direction.
      > So it is horses for courses, if you want to paint your robot blue then that is ok but it isn't necessary for the most part.
      > Sure some fancy stuff can be done if you get the control right.
      > But PID control isn't going to make a poorly designed robot work really well.
      > And a poorly implemented PID controller can make a robot, which would otherwise work fine, work really poorly!
      >
      > Now if there were a whole lot of SR04 robots about, which there aren't but by now there ought to be, which had all the sensors and uC but could be improved by implementing PID control then it would be time for some PID tutorials and code samples and general help to get the robots working better. But the robots don't exist! And my view is that a lot of people make out that you need complex mathematics (not really but a lot of people would see it that way) to make them work. And so people don't make the robots because they just KNOW it is going to be too hard or too much work or just too.... So what do we get instead - one (or two) motor bristle bots - hey look at the little bugs go!
      >
      > DAvid
      >
      >
      > ----- Original Message -----
      > From: dpa_io
      > To: SeattleRobotics@yahoogroups.com
      > Sent: Saturday, February 05, 2011 2:21 AM
      > Subject: [SeattleRobotics] Re: SR04
      >
      >
      >
      > Hi David,
      >
      > There are some other advantages to using PID control aside from the ones you've mentioned, like:
      >
      > 1. Keep the robot running in a straight line when the wheels are on two different surfaces. (assuming differentially steered robot). For example, one wheel on a high friction surface like carpet and the other on a low friction surface like hardwood floor.
      >
      > 2. Allow the robot to drive at extremely low speeds. Well below the torque requirements of platform itself. Larry Barello showed me one of his robots moving inches per hour. This might seem trivial but with a differentially steered robot, it translates into the ability to steer very precisely, as the inside wheel in a turn is often turning very slowly.
      >
      > 3. Make robot behavior independent of battery voltage. That is, the behaviors remain consistent as the battery drains.
      >
      > 4. Make the robot behavior independent of the surface on which it is running. This turns out to be particularly useful if the robot is not using odometry. So that turns, for example, can be decoupled from the friction of the particular surface the robot is running on.
      >
      > cheers!
      > dpa
      >
      > --- In SeattleRobotics@yahoogroups.com, "David Buckley" <david@> wrote:
      > >
      > > Randy
      > > I can feel for dpa. Years ago when I first had my robots at University organised functions I got "oh but they are made of wood! WE use Lego and Poster putty" [said in all seriousness :-( ].
      > >
      > > However.
      > > Used right, motors don't need PID for a robot to function REALLY well.
      > > http://www.davidbuckley.net/DB/MM3drawing1.htm
      > > shows the first drawing made in 1982 by my third computer controlled robot. The motors were just turned on and off by relays controlled by timing loops in a simple ZX-81 program.
      > > (Later I had a much more complicated program, not on the web, which allowed routines, written in robot code, to be strung together)
      > > [finding the URL I discovered I had had a case error (k instead of K) since 2006! Ooops, thank you.]
      > >
      > > Using vinyl rubber tires on paper that is about as good as you can get.
      > > Running Parallax's BoeBot with modified continuous rotation servos it is possible to get equivalent performance.
      > > My Zero2 Turtle http://www.davidbuckley.net/DB/Zero2.htm using a precision aluminium chassis and stepper motors and neoprene tyres could not better it because of tire slippage on the paper and the slight drag of the umbilical.
      > >
      > > You now know why I wasn't impressed by the tests of the SRS robot on that web page mentioned before. I am sure the poor results were nothing to do with the SRS robot, just the software used for the tests, which of course wasn't shown.
      > >
      > > PID from shaft encoders is necessary where you want real zippy performance such as in fast micromice, but for the average wheeled robot it is a waste of time and energy because of wheel slippage on the floor/table/whatever.
      > > If you are using GPS then you know where you are and where you want to be and it doesn't matter a bit how the wheels slip.
      > > If you are seeking light or some other beacon then that is what you aim for and it doesn't matter a bit how the wheels slip.
      > > If you are trying to do odometry with the expectation that your robot will be where it thinks it is then you would be better off taking up fishing. PID or no PID.
      > >
      > > DAvid
      > >
      > >
      > >
      > > ----- Original Message -----
      > > From: Randy M. Dumse
      > > To: SeattleRobotics@yahoogroups.com
      > > Sent: Friday, February 04, 2011 7:50 PM
      > > Subject: RE: [SeattleRobotics] SR04
      > >
      > >
      > >
      > > David Buckley said: Wednesday, February 02, 2011 4:55 PM
      > > > No wonder SR04 was/is held in awe.
      > >
      > > Yes, exactly. I think SR04 is the base line for what we should
      > > require in a differential drive robot. Working PID and Odometry
      > > and basic navigation to a point are in the reach of every
      > > hobbiest. RC servos are okay for a first experience, but the
      > > equivalent of adding wheel watchers, or encoded gear motors
      > > should be the very next step.
      > >
      > > Like I say, seems the Scribbler 2 is taking that step, and will
      > > be a much better robot base for it.
      > >
      > > I appeared on the bobbiest scene just as SR04 was winning yet
      > > another CANCAN contest at DPRG. (I'd been doing stepper and DC
      > > Servo controls for something like 20 years before that.) So to
      > > me, SR04 was the functionality to beat. I never knew dpa got
      > > grief and negative feedback for his contribution. My sense was
      > > he was worshipped. So this discussion has given me a new insight
      > > on dpa's travails.
      > >
      > > Randy
      > >
      >
    • dpa_io
      Sigh. Best luck to you David. dpa
      Message 69 of 69 , Feb 8 8:10 AM
      • 0 Attachment
        Sigh. Best luck to you David.

        dpa




        --- In SeattleRobotics@yahoogroups.com, "David Buckley" <david@...> wrote:
        >
        > David
        > Just what are you trying to say?
        > You are just mixing everything up.
        > You talk about brushless motors and ESCs - then when I mention a possible application of those motors, ie the motors which are used for driving propellers which spin fast, you then flim flam and then tell me that brushless motors are used in some servos as if it is some sort of secret that only the cognoscenti know. Well quite a few people know that, however those motors have their own ESC.embedded in the servo rather than using a separate module. Its like talking about apples and rocks.
        >
        > The brushless motors used for aircraft propulsion are high power for size and low duty cycle - as you know. Would they have been suitable for SR04 - not really.
        >
        > I am sure you can adapt motors to gearboxes, so can other people including me, but many can't because of skill or facilities.
        > As for building experimental robots, you have built a few and so have some of the rest of us.
        >
        > It is good that some people can do all three of mechanical, hardware and software otherwise every robot would end up being the proverbial designed by committee mishmash. However as for getting together with some mechanically inclined friends, chance would be a fine thing, I am the one they turn to when anything needs fixing or making whether it be mechanics, electrics, electronics, computers, software, woodwork, plumbing and you name it. On the other hand I listen to them playing music and singing and telling jokes and stories. Maybe you are lucky to have friends who can do practical things better than you.
        >
        > Then its more flim flam regarding GPS modules, it's back to odometry.
        > Then more smoke screen when I ask how the r/c-radio can be adapted to robotics, not a mention of servos there.
        >
        > As for paying to see your hand, well! You obviously missed the point that I wasn't asking you to tell me, I was asking so you could give the benefit of your wisdom and expertise to the other members of this group.
        > If you can't, from your personal experience of the new technology explain to this group how they might use it then what is it but being bedazzled by it.
        >
        > I don't know why you are getting so hot under the collar, was it the bit about the papers?
        >
        > DAvid
        >
        >
        >
        >
        >
        > ----- Original Message -----
        > From: dpa_io
        > To: SeattleRobotics@yahoogroups.com
        > Sent: Tuesday, February 08, 2011 6:03 AM
        > Subject: [SeattleRobotics] heli electronics (was Re: SR04 )
        >
        >
        >
        > Hey David,
        >
        > see insert comments.
        >
        > --- In SeattleRobotics@yahoogroups.com, "David Buckley" <david@> wrote:
        > >
        > > There is one flaw in all this.
        > > As you mentioned your motors were from copiers.
        >
        > Some of them, I've used all type.
        >
        > > The rc world outrunner brushless motors even though the redesign reduces speed and increases torque still run too fast and need a gearbox for most robotic applications. Ok some have maybe ~3:1 gearing or belt drive reduction, but are still too fast without a gearbox. Trouble is there aren't any! You would have to build your own.
        >
        > Your point? ;>)
        >
        > > The power output and current draw of outrunners is rather higher than is appropriate for most hobby robots.
        >
        > Define "appropriate" if you would.
        >
        > > Hence the rc world is not the right place to look for drive units, so it's back to scavenged motors WITH gearboxes.
        >
        > I've adapted several gear boxes to motors that came without them. It's generally not that hard.
        >
        > > All the rc brushless ESCs are designed for these fast powerful motors.
        > > So how do you overcome that problem?
        >
        > Hmmm... Well, its is true that building experimental robots requires some, ahhhhhh.... robot building. If you don't want to do that part then maybe your not that into, you know, building experimental robots.
        >
        > Maybe a team effort with some mechanically inclined friends, the ones who'd like to build a robot but who are not interested in writing code.
        >
        > The thing that has always fascinated me about robot building is that it is really a combination of 3 skills: mechanical, hardware, and software. Like many in the hobby I like to do all three. YMMV
        >
        > >
        > > For a legged robot using linear actuators, brushless motors driving backdrivable lead screws would be about right. Super high power in short bursts. But the cost! You really need at least six servos per leg for a humanoid, ie 12 servos, or 3 servos per leg for a quadruped, again 12 servos. So 12 outrunners + 12 ESCs + 12 custom made drive units at the minimum. Not your average hobby project. It is immediately 12 times more expensive and 12 times more complicated than an rc helicopter and you still don't have the gearboxes!
        > > Better off using Dynamixels.
        >
        > Maybe some misinformation here, David. There are lots of R/C servos available today that use brushless motors. I have them on my helicopters. Google it.
        >
        > >
        > > Now as we have just seen the DIYdrone people have been instrumental in the creation of small cheap GPS modules. Resolution 3m, so only good for far ranging outdoor robots, great for asking "Am I downtown yet?", not so good for "Is this the kitchen or the living room?".
        >
        > Well, an easy way to do that is to use odometry. Even poor odometry is good enough to tell you what room you're in. I get the impression from other comments you have made that you think unless sensing is perfect that it's not useable. In my experience that is not even remotely true. Almost the opposite. Again, in my experience, even poor sensors can produce robust reliable results. But you have to be clever. Don't get caught in the "I can't do what I need to do because the sensors are not good enough" trap.
        >
        > > The radio control systems are designed for human control of something. I want my robots to be in control of themselves while doing a task I specify. How can the radio control systems be used for robotics and not for controlling something which is called a robot but is really another rc model?
        >
        > What? You're using R/C servos now. You already know how to control R/C systems from a computer. I assume. If not, plenty of people on this list can assist you.
        >
        > >
        > > Your ultrasonic autofocus units were off the shelf, mass market, but now there are much better units and cheaper developed specifically for the hobby robotics market. obviously a big enough small mass market.
        >
        > This is my point, which I won't belabor further. We (robot builders) take stuff developed for other disciplines (sensors, actuators, microprocessors,et al) and adapt them for our purposes. That's the nature of the beast. R/C servos are a perfect example.
        >
        > >
        > > How do you see robotics using the latest rc high tech stuff?
        >
        > Oh David, you have to pay to see that hand... :>}
        >
        > >
        > > On the other hand there is stuff out there just waiting to be used.
        > > I have just bought some cheap £10 12v battery drills. Motor + gearbox + rechargeable battery for less than I can buy a rechargeable battery. High tech, high volume, low price.
        > >
        > > Apparently there are defunct (cheap) Rumbas on eBay with working drive units.
        > >
        > > Thing is not to get bedazzled by the green grass on the other side of the river, nor the methods and equipment we use at work, nor the neat but expensive stuff for another hobby.
        >
        > Whoa! That's a little bit rude. Could be that what you see as "bedazzled by," others might see as "informed by," or even "inspired by." Chacun a son goût
        >
        > "Thing is" not to be so myopic as to miss the opportunities around you.
        >
        > best,
        > dpa
        >
        > >
        > >
        > >
        > > ----- Original Message -----
        > > From: dpa_io
        > > To: SeattleRobotics@yahoogroups.com
        > > Sent: Monday, February 07, 2011 4:21 PM
        > > Subject: [SeattleRobotics] heli electronics (was Re: SR04 )
        > >
        > >
        > >
        > > Exactly, Dan! What I'm suggesting is that we can piggy-back off their advances. As we've always done.
        > >
        > > The sonar on my robots were developed for auto-focus cameras, not hobby robotics. The IR detectors for TV remotes, the motors for copier machines, the motion detectors for burglar alarms, and on and on.
        > >
        > > That was all in the last century. Now we have off the shelf ESCs with built-in PID controllers, IMUs, brushless motors, etc, etc, but they don't seem to have penetrated into hobby robotics. We seem to be stuck in the last century. But there is no reason to be that I can see.
        > >
        > >
        >
      Your message has been successfully submitted and would be delivered to recipients shortly.