Lynxmotion Tech Support

Hi,

For a long time I've been thinking (like many others) about adding sensors to my hexapods for TA (terrain adaption). This might be a waste of time since Jim already has proven to make some excellent microswitch foot sensors.

Anyway, my idea was to use FSR sensors between SES brackets. Instead of having the sensor/switch on the foot I wanted to try adding a sensor in other parts of the leg, making the whole leg "ground sensing".

My first attempt was to add a sensor inside the femur part, this is something I tried for about a year ago:




I'm using a 3mm screw for adjustment.


Using a little tape sticker for protecting the FSR


To make a long story short, I wasn't satisfied with the result.. And I didn't like the idea of drilling new holes in the SES brackets either.

But I didn't wan't to give up the FSR yet..

This time I thought of placing the FSR inside the coxa, or between the coxa and femur bracket to be more correct. Yesterday I made two very simple parts of 1,5mm aluminium. The parts are mounted to the ASB-04's using the small counter sink screw that came with a pair of the ARF-01's (SES feet) (I wish I had more of those screws.. ):


As you can see the plate without the FSR is a little shorter to make space for the rubber bumper pad. The two plates are mounted together with two rubber parts. Guess where they come from..



Pardon the shiny metal, I didn't got time for sandblasting. After all this is just a prototype for testing.

Optionally, I made space for a spring between the brackets. At first I didn't use the spring since the weight of the leg will force the brackets apart anyway.


Closeup of the rubber bumper mounted:


The two plates are mounted together with two screws and the rubber parts I mentioned:


Some pictures of the final result:


I'm using a longer screw between the brackets (where I mentioned the spring). Adjust this screw just enough to make a little gap between the FSR and the rubber bumper pad:


From another angle (the coxa side):


And from the femur side:


Example picture with servos:


I've done some small testing and it seem to work perfect so far. I also need to do some more testing on a robot to see if it fails under certain conditions. I'm still not convinced if this solution work as good as an actual foot sensor. Anyway, I want to try this on my T-hex. I just need to get more of those small counter sink screws..

Any comments are appreciated.

_________________
[b]Kåre Halvorsen, Zenta[/b]
-----------------------------------------
Zenta's blog
http://zentasrobots.com/
Zenta's YouTube channel
http://www.youtube.com/ZentaOlbaid
-----------------------------------------

I'm not surprised the first trial didn't show much either.

Looks like a 2nd sensor at the opposite end of the plates would be useful. Or even three sensors arranged in a triangle. Then you'd see all the forces, either leg loaded, or unloaded.

You almost need to be a structural engineer...

Alan KM6VV

_________________
Visit:
http://groups.yahoo.com/group/SherlineCNC/
http://tech.groups.yahoo.com/group/HexapodRobotIK/

I believe that can easily be very complicated to achieve. It's the down-force I wan't to detect. The main idea was that the two plates should work as a hinge with the sensor at the other end. I hope to do some more testing on a robot soon to see if this is a realistic way to go.

_________________
[b]Kåre Halvorsen, Zenta[/b]
-----------------------------------------
Zenta's blog
http://zentasrobots.com/
Zenta's YouTube channel
http://www.youtube.com/ZentaOlbaid
-----------------------------------------

Sounds like a fun simple way to try out. It will be interesting to see the differences in force readings when the leg hits at different angles and the like.

But I am not sure about those shiny pieces

Kurt

That looks great!

Is there much deflection when the leg is lifted?

Oh nevermind, I see you have a screw limiting how far it can deflect. Great idea!

Have you thought about adding sensors to the coxa? I haven't figured out a way after thinking about it for a while. Possibly on a T-rex, where C-brackets are used (if I remember right) to support the coxa servo. It would probably boil down to a bracket similar to what you've come up with. Very clever!

Alan KM6VV

_________________
Visit:
http://groups.yahoo.com/group/SherlineCNC/
http://tech.groups.yahoo.com/group/HexapodRobotIK/

Exactly, there is very little deflection due to the screw you mention. I think the best way is to use some locktite on the screw/nut to keep this distance solid. Also, I believe the deflection caused by the gear slope/backlash from the femur servo give much more deflection, so i don't think that will be a problem.

_________________
[b]Kåre Halvorsen, Zenta[/b]
-----------------------------------------
Zenta's blog
http://zentasrobots.com/
Zenta's YouTube channel
http://www.youtube.com/ZentaOlbaid
-----------------------------------------

I'm planning to only use the FSR as a switch (1/0), not read the actual force. The main theory is that this sensor should work at any foot angle. It only remains to see if this theory work in the real world..

_________________
[b]Kåre Halvorsen, Zenta[/b]
-----------------------------------------
Zenta's blog
http://zentasrobots.com/
Zenta's YouTube channel
http://www.youtube.com/ZentaOlbaid
-----------------------------------------

Note to self... Send Kåre a handful of counter sink 2-56 screws. Anything else you need let me know right away.

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

So you got the hint? Thanks Jim, I'll let you know. I'll soon post something very interesting when it comes to the 2DOF SES based joint for ankle and hip on a biped.

_________________
[b]Kåre Halvorsen, Zenta[/b]
-----------------------------------------
Zenta's blog
http://zentasrobots.com/
Zenta's YouTube channel
http://www.youtube.com/ZentaOlbaid
-----------------------------------------

As usual, great idea, great workmanship, and great post. Looks like a nice way to make a contact switch. For longevity, loc-tite might be good for the "hinge" screws because of the movement in this area. That's about all I can think of. Great job and thanks for sharing.

By the way, I like those shiny prototype pieces.

_________________
Mike

The robotics nut.