Alt-BEAM Archive
Message #08819
To: beam@sgiblab.sgi.com
From: "Timothy Flytcher" flytch@hotmail.com
Date: Mon, 03 Jan 2000 22:51:35 PST
Subject: [alt-beam] Re: BI ped
>From: Evaristo Westplate
>I have been working on a biped robot about four years ago as a study.
>I have a photo somewhere of the robot. If someone is interested
A picture is worth a thousand words...
I would love to see a pic... :)
Timothy...
______________________________________________________
8820 Mon, 03 Jan 2000 22:58:31 -0700 [alt-beam] Re: BI ped beam@sgiblab.sgi.com Senior I'd really like to know what you are using for pressure sensors. What is
that status of your bot, Ian?
Your idea, Evaristo, whether it's original or not, is ingenious! The
more pressure on one side, turn the ankle so it exerts more pressure to
balance it out. You's have to reverse this when putting the foot down,
so that the more pressure on a side, the less pressurre the ankle puts
on that side, and the whole thing about the ankle conturing to the
ground is solved!
The only problem I see is that this will keep the leg attached to the
ankle going straight up, which we might not always want. You'd have to
have a way of biasing the ankle to get it to bend so the bot can walk.
Do you get what I'm saying?
Kyle
Evaristo Westplate wrote:
>
> Hi,
>
> >Yes, please send a link to the picture!
>
> I will look for the picture and setup a web page for it.
>
> >Just curious... could you explain the pressure sensors a bit more?
>
> Each foot has four pressure sensors. One in the "toe", one on the "heel"
> and the two others on both sides (sensors are spread like a cross).
> The toe and heel sensors are linked so that they form a voltage divider.
> When both sensors receive the same amount of pressure then the output
> voltage will be roughly 0.5*Vdd. Same goes for the other two sensors. This
> setup will work regardless how heavy the robot will be (taking into account
> the maximum range of the pressure sensors). When the robots leans forward
> then the front sensor receives more pressure while the back sensor sees a
> drop in pressure.
>
> This setup was tested first with an ankle joint and a foot turned upside
> down. We placed a bottle on the foot and the controller kept the bottle
> upright. Very cool when the leg (with ankle and foot attached) is moved
> around and the bottle stays upright. The pressure sensors work very well
> and small changes in weight distribution can be detected very accurately.
> Ofcours this setup will only work if the upper body of the robot can't move
> otherwise you will need something like a gyroscope in the upper body to
> respond to moves of arms or something else. The robot we build used mostly
> the ankles to stay upright. Once you start moving it becomes a different story.
>
> Hope this helps. As soon as I have the photo up I will post the link.
>
> See'ya
>
>
> Evaristo
>
> Gizmo homepage: http://www.crosswinds.net/~evaristo
8821 Tue, 04 Jan 2000 15:27:53 +0100 [alt-beam] Re: BI ped beam@sgiblab.sgi.com Evaristo Westplate Hi,
The pressure sensor used is an FSR 150.
FSR sensor:
Manufacturer: International Electronic & Engineering
FSR models (partial list): FSR 149, FSR 150, FSR 151, FSR 174
Manufacturer web site: http://www.iee.lu/
If you look at the pictures of the biped robot you can see narrow black
strips with two white lines coming from beneath the foot and going to a
circuit board. In Holland the sensor is available through Conrad. Don't
know about the rest of the world.
Pictures of the robot can be seen at:
http://www.crosswinds.net/~evaristo/biped.html
>Couldn't this foot with four pressure sensors be done with a basic 3DOF head
>circuit? I mean, a head simply reacts to the difference in resistance of
>each photodiode. Could you replace those photodiodes with pressure sensors
>and basically make a 3DOF ankle?
Yes it could be done but when using two feet you need some mechanism to
overwrite it and position the servos attached to the foot. The problem is
that when the left toe feels to much pressure, it will compensate for it.
As a result, the right heel will sense to much pressure and will compensate
for it. As a result the left toe will sense to much pressure and will
compensate for it. As a result the right heel....... etc.
We solved this problem by letting one foot be dominant. For a beam solution
something new has to be made.
>The only problem I see is that this will keep the leg attached to the
>ankle going straight up, which we might not always want. You'd have to
>have a way of biasing the ankle to get it to bend so the bot can walk.
>Do you get what I'm saying?
The ankle is build using two servos so it can move in two directions. This
also includes bending the lower leg. Have a look at the pictures in the
above mentioned link. I hope this is what you mean otherwise let me know.
See'ya
Evaristo
Gizmo homepage: http://www.crosswinds.net/~evaristo
Home