Putting robotics at your service™

Free shipping on orders over $75

Ground contact sensors

Print view Share :
Previous topicNext topic

Page 1 of 6 [ 86 posts ]

1, 2, 3, 4, 5, 6
Expert ( offline )
Posts: 471
Posted: 2006-02-21 14:19 
 Post subject: Ground contact sensors
I just received delivery of a bunch of FSRs which despite the documentation, are extremely sensitive at low pressure.

I've also got a bunch of these ir sensors.

So I need to figure out some way to physically mount them on the legs. The FSRs really like to be compressed between to parallel plates and I'd like reasonably consistent readings when the foot is placed down at an angle rather than absolutely vertical.

I'm thinking maybe half a gumball bouncy ball replacement foot, but how to mount it in a manner that will allow the motion to depress the sensor? I figure the IR sensor might just hang off the side someplace.

Any ideas would be appreciated!


User avatar
Guru ( offline )
Posts: 9256
Posted: 2006-02-21 15:29 
andylippitt wrote:
So I need to figure out some way to physically mount them on the legs. The FSRs really like to be compressed between to parallel plates and I'd like reasonably consistent readings when the foot is placed down at an angle rather than absolutely vertical.

I'm thinking maybe half a gumball bouncy ball replacement foot, but how to mount it in a manner that will allow the motion to depress the sensor? I figure the IR sensor might just hang off the side someplace.

Any ideas would be appreciated!

I have been working on this for way too long. I have some ideas, but nothing that has come to fruition yet. I'm referring to the tubing leg used in the servo erector set, but I am open to feedback on any type of design anyone comes up with. I may do it all in aluminum with a large ball joint above a 1.25" disk (foot pad). This aproach may cost $20.00 to $30.00 for each foot! Bots need feets! :D

_________________
Jim Frye, the Robot Guy
http://www.lynxmotion.com
I've always tried to do my best...


User avatar
Guru ( offline )
Posts: 3585
Posted: 2006-02-21 15:45 
Here is one idea I came up with. Without having the sensor and the hex sitting in my lap, this is the best I could come up with using just pictures.

There could also be a spring attached to naturally hold the sliding "U" bracket out, releiving pressure on the sensor contact.


_________________
Mike

The robotics nut.


Veteran ( offline )
Posts: 166
Posted: 2006-02-21 16:01 
andy and i had an IM discussion, here is a couple of things i shared with him:

the quad i built some years ago used the universal joints out of an rc car to attach a small pad to the bottome of the foot. then a length of fuel tubing was stretched over the universal to hold it relatively stiff. it becomes kind of a passive ankle.

at the time i had not seen these sensors so i never tried but i did get substantually more grip and a sort of shock effect that smoothed the ride a bit. later on i added a spring on the outside when the bot became a little over weight.

like this, but all plastic http://www2.towerhobbies.com/cgi-bin/wt ... LXGE52&P=7 i got a set of four in a bag for like 20 bucks

at one time i saw a hex with notch cut out of the final leg member and the force sensor inserted here, that way when it bent in respnse to a force applied at the tip, the sensor recieved a portion of that load. as i remember the info was not useful inregards to absolute force, something about to many variable to calculate, but in relative force...

just thought id share
nick


User avatar
Guru ( offline )
Posts: 3585
Posted: 2006-02-21 16:38 
Here is a rendered image of what the bracket may look like for the sensors.


_________________
Mike

The robotics nut.


Rookie ( offline )
Posts: 21
Posted: 2006-02-21 17:09 
I've started construction of my quad (used to be a radially symmetrical hex but it's been scaled down due to costs) and wanted it to sense the floor - or lack of it! I'll be using QTC, quantum tunnelling compound, either cutting out a notch in the bottom of the foot to put a pill in direct contact with the floor, or cutting into the side of the leg to form a hinge in which the pill is held.
I live in the UK so the only place I've seen that carries QTC is Maplin, but I'm sure you'll be able to find it elsewhere.
http://www.maplin.co.uk/Module.aspx?Mod ... c&doy=21m2
Hope this helps,
Dave


Expert ( offline )
Posts: 471
Posted: 2006-02-21 18:06 
Jim, a slightly different issue I has considered was placing two on each joint. It seems to me that a notch coming off a servo horn could align with a wishbone style notch on a bracket to provide it's torque. Between the horn-notch and the sides of the wishbone could lie two of the FSRs that would give good torque data. How to do all this without compromising the strength of the joint, I don't know. I'm a programmer, not an engineer :)


Guru ( offline )
Posts: 1660
Posted: 2006-02-21 18:48 
I'm not exactly sure if this would work for you, but...
What about sticking a slightly smaller diameter tube inside the Lynxmotion tube?

If you put a metal plate inside the lynxmotion tube to stop the tube that's sliding in, and put a flat cap on the end of the sliding tube, you could place an FSR between the two plates.

To keep the tube from falling out, or the FSR from being compressed too much, you can drill a hole through both tubings, and stick a pin/screw/nail/whatever through that hole.
Then you would widen the hole on the outter tube until it was a vertical slot that stopped the inner tube from sliding too far, either way.

_________________
Disclaimer: The above person does not claim to have knowledge pertaining to the following subject: anything. Thus, said person may not be held liable for any mishaps/explosions that his advice incurs.


Expert ( offline )
Posts: 471
Posted: 2006-02-21 18:49 
OK, this is far from robustly tested, but I got ultra low tech and I seem to have pretty good results.

http://vizlog.com/robot/GroundContactSensor.avi

I removed the rubber foot. Taped the sensor to the flat edge of the three lexan pieces at the bottom. Rummaged around and found what is probably perfect, if a bit silly. I have a TO-92 package transistor (cylindrical with one flat side) placed such that the flat edge is resting on the sensor. The rounded side couples with the inside of the rubber foot. Slide the whole thing back on, adjusting it so that the it's attached just shy of registering on the sensor.

The sensitivity it quite good and the amount of friction between the foot and the leg is high enough that I seriously doubt slippage will be a problem. Angled contact seems to be quite good too.


Expert ( offline )
Posts: 471
Posted: 2006-02-21 18:50 
Just in case it wasn't clear, the transistor isn't included in the circuit, I'm just using the physical space it takes up to couple between the rounded rubber inside of the foot and the flat sensor.


User avatar
Guru ( offline )
Posts: 3585
Posted: 2006-02-21 19:00 
Sounds like you have a good way to install these sensors. Another test would be how well will the assembly hold up, or will the readings begin to go south due to shifting.

_________________
Mike

The robotics nut.


Expert ( offline )
Posts: 471
Posted: 2006-02-22 00:55 
I'm not viewing this as a done deal. In fact, I don't really like tape and relying on rubber on plastic friction from keeping these from giving me consistent readings. I'm still wide open to ideas!


Rookie ( offline )
Posts: 42
Posted: 2006-02-22 06:52 
Hello!

what about a light sensitive sensor mounted inside the aluminium tube of a servo erector set? you have to skip the rubber endings, though and dirt may be a problem
When the "foot" is placed on the ground, everything goes dark, and voila, you can read "ground contact"...
Or maybe you can do it just like those optical mouses: one LED per leg and the sensor from the mouse mounted inside the tube.

"it is just one small step for a hexapod, but a giant leap for its creator" :)

_________________
Marc

"But I just keep on laughing/Hiding the tears in my eyes"

[Taken from "The Cure" famous "Boys Don't Cry"]


Expert ( offline )
Posts: 430
Posted: 2006-02-22 11:08 
I haven't done it yet, but my plan is to try an IRPD arrangement.
But, not using the Sharp type of modules, but rather using a Hamamatsu S6986, as seen here: http://www.junun.org/MarkIII/Info.jsp?item=79 .
This little guy drives an LED with a certain pulse rate, and detects the 'bounce' of that same signal, and produces an analog output related to how much bounce there is.

Pete

_________________
Pete in Colo. Springs


User avatar
Guru ( offline )
Posts: 3585
Posted: 2006-02-22 13:29 
Now thats a neat little device :lol:

_________________
Mike

The robotics nut.


1, 2, 3, 4, 5, 6

All times are UTC - 5 hours [ DST ]. It is currently 2018-05-25 02:37