Subject:
|
Re: WWW spider
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Mon, 8 Dec 2003 15:57:19 GMT
|
Viewed:
|
1360 times
|
| |
|
|
In lugnet.robotics, Paul Kleniewski wrote:
>
> "Kevin L. Clague" <kevin_clague@yahoo.com> wrote in message news:HpDyM3.A0z@lugnet.com...
>
> > Yes, I understand that you would have the downward pressure of the step move
> > stop the leg movement by cutting off pressure to the piston(s) driving the foot
> > down. Once this pressure is cut off by the switch, presumably the other legs
> > continue with the sequence.
>
> yes
> i tried some sequensers during weekend
> and i'm on the way to create this functionality
> i hope it will work :)))
Good luck! I look forward to it.
>
> > Eventually the sequence gets back to the foot that is on the step and the foot
> > does not move (because the switch, acting as a touch sensor) still has the
> > pressure turned off). It seems like the leg never starts moving again.
>
> i think this could be done by the mechanism which could work as "clear" method
> so let's say at the end of the sequence there will be a step called clear
> and this step will set/reset some of the "dead" states
>
> pistons which will akt as clear
> should be connected to valves not in the same way as "normal" pistons
> i assume "normal" piston is connected to valve let's say 1:1
> so if the piston moves in or out the valve is moved from one end to the other end
> but the clear should work in the way which would move the valve to the only one direction
> when clear piston gets back it doesn't move the valve
>
> of course the meaning of "clear" depends on you
Hmmmmm.... I wondered if you might be thinking of this.
A common method used in the past was to use a central pneumatic timer circuit
that control the muscle pistons and make them move. This is what Doug Carlson
did in his beatiful and ground breaking pneumatic hexapod:
http://www.visi.com/~dc/hexapod/index.htm
The central timing circuit feeds the muschle pistons in the hips and legs. One
downside to this is that the timing circuit and its piston switching is
independent of the muscle expansion/contraction rate of the legs and hips. Doug
had to make sure that the load on the timing pistons was greater than the load
on the legs or hips. This way Doug made sure that the legs and hips always hit
their expand/contract limits. The downside to this is that the heavy load on
the timing circuit means his hexapod had to have high pressure to walk.
I tried using this technique on some designs, but didn't like the result.
My strategy for making sure the pistons hit their limits is to make the muscle
pistons also play the role of timing pistons. Quad 242 has a total of 10
pistons. Eight are both muscle and timing pistons (which is why they all have
switches used to instrument their state). Only two of the pistons are timing
only.
When you "clear" the pistons to create alternate behavior you'll have to decide
whether you really want to change the muscle/timing system's state, or simply
ignore the switch outputs of the hung leg and instead use alternat values to
pass into the state machine. It sounds like the second plan might be
inexpensive.
Expanding my sequencer design methodology to full blown state machines will be
interesting.
>
> > How do we tell the difference between the switch/pressure sensor being flipped
> > because of premature end of step and flipped because it is on the floor?
>
> in these terms i think there is no difference between
> "on the step" and "on the ground"
>
> > I'm interested in your idea, but I'm not sure how to use it. It may be me
> > because I've not progressed from simple sequencers to complete pneumatic state
> > machines that change state because of external inputs as well as current state.
> > Quad242 plus the polarity reversers can do this if you consider the positioning
> > of the polarity reversers' handles as inputs.
>
> i thought about the sequencer as about electronic machine :)
Oh...
>
> > I suppose that if you combined the switch that is affected by weight bearing or
> > not in series with a switch that tells you if the leg is completely extended or
> > not, you could tell if the leg hit the ground prematurely.
>
> yeah!!!
> it's something like this
>
> > Do you really want to leg to freeze once we detect this situation? This is
> > certainly not what my dog would do. He'd extend the leg and keep going. It
> > must be something special about the WWW spider that its trying to keep the body
> > at the same altitude no matter what the terrain?
>
> no no no
> WWW spider has nothing to it
> it was just the idea to put additional elements into the circuit
> which could "inform" the system there were some changes in the environment
>
> "to inform" means to robot "do something extraordinary"
> something different that the primary sequence
> and get back to the original sequence if the cause will disappear
Ah yes. When I refer to state machines I mean exactly this, where the pneumatic
circuit has a current state and inputs. From this it calculates a new state. A
sequencer only has current state. From the current state it calculates the next
state (no matter what the outside conditions).
>
> > I think I understand your idea, I'm just curious how it fits in the big picture
> > (remember I'm trying not to have the RCX control articulation).
>
> sure!!! i know
> the same is with my new hexapod
> i want to do all movements from one motor which will rotate in one direction only :)
Will you have two groups of three feet, where the two feet groups are 180
degrees out of phase?
>
> br
> pixel
Kevin
|
|
Message has 1 Reply:
 | | Re: WWW spider
|
| hi i just developed this sequencer DL (URL) there is no purpose for this mechnism :) it shows 2 things 1. even small and simple changes in the circuit make it a little bit complicated :))) 2. it can be done :) it's just a first trial i'll work on (...) (21 years ago, 9-Dec-03, to lugnet.robotics)
|
Message is in Reply To:
| | Re: WWW spider
|
| "Kevin L. Clague" <kevin_clague@yahoo.com> wrote in message news:HpDyM3.A0z@lugnet.com... (...) yes i tried some sequensers during weekend and i'm on the way to create this functionality i hope it will work :))) (...) i think this could be done by (...) (21 years ago, 8-Dec-03, to lugnet.robotics)
|
60 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
