Hexapod Kit Design: MSR-H01

[Lazer155]

Thanks, for that diagram. I understand what everyone was talking about with the 2 different range finding sensors now. I could not figure out why people kept mentioning the SRf04 and some would say to use the SRF08 sensor. Now I understand though.

[EdG]

Hello Forum!

When the MSR-H01 is in it's normal standing pose, are the servos actively supporting it?
Or, to put it another way, would (somehow) measuring the current in a servo's motor give an indication whether that leg was supporting the hexapod, or raised in the air?
You might be able to guess where I'm going with this.

I'm not that familiar with servos - I missed out on the whole R/C thing, and I don't have an MSR-H01 to try this out with *yet*, but I'm guessing the answer is yes.
I just think I might be disappointed if I couldn't at least have a go at detecting varying terrain, so please dispel my feeble excuses not to get one.

[Matt Denton]

Hi,

Yes the servos will be taking a current draw in the standing position, so you can monitor the current through the motor to detect torque, but you will need to modify the servos.

[marcel.kanter]

Hello,
you need not to modify the servo itself. You can add an shunt in the supply lines of the servo and measure the current drawn by the servo. In supporting position it should draw more current, since it has to generate the momentum/force to hold the position nad the load.
What you have to take into account is that the servo electronics (the controler itself) consumes power itself.

If i'm guessing correctly you're trying to get ground adaption without additional sensors! Btw. i'm trying the same with my servoctrlmini (which has the current sensing amplifier on board). My plan is to feed the controller with the current directly. So that you first send an command to move to an position (with regular position based control) and then switch to current control and the servo regulates the current (and with this the force by the motor). But this is a plan and has to be evaluated.

Link to the servoctrlmini page on my hp: http://www.marcel-kanter.de/projekte/servoctrlmini (the project is under development)

[EdG]

Thanks for the rapid reply Matt. Sorry if I posted this in the wrong place, by the way.
I failed to resist ordering a kit anyway. Thanks for making a 'Want one!' item into a 'Can have one!' with this kit.

Marcel - that looks very interesting indeed. It would be nice to have total control & feedback. I guess direct control of current also opens up some very nice control possibilities. I hope the project is going well.
It sounds hopeful to investigate the current measurement route for terrain detection, though I imagine other approaches might be more successful. I just like the idea of making the most of existing parts. Maybe all it would need is measuring the motor voltage? Something to look at if I get the time. (So many toys, so little time etc.)

Great forum & nice to see lots of active & varied ideas here.

[john-harper]

I've been thinking about the same thing, and looking at various ways of doing it. For the "proprioception" approach (i.e. measuring what the servos are doing), my thinking is to use Openservos. They have a whole bunch of monitoring functions including the actual motor current. The only disadvantage that I can see is that you have to modify all of the servos. I have the bits to make one of them and plan to do some tests in a test rig to see how well it really all works out.

[marcel.kanter]

Ok, controlling with current as reference variable works. I've adapted the existing position-pid implementation to an current-i controller for this task and the behaviour is as expected. The servo horn moves in one direction as long as the current is not reached. If it is reached it stops, if you push the horn the pwm is reduced to 0 and there is only friction by the gear set. This can be eliminated by pairing the current controller with an near zero friction type (i've already tested too).
(I'll report the further development on my homepage and maybe in an extra thread in the servo section of the forum)

[john-harper]

When the MSR-H01 is in it's normal standing pose, are the servos actively supporting it?

Yes... after all, Newtonian physics says SOMETHING must be. Depending on the posture there's a bit of math involved in figuring out exactly which servo is doing what, but it's all there.

Or, to put it another way, would (somehow) measuring the current in a servo's motor give an indication whether that leg was supporting the hexapod, or raised in the air?

Maybe... that's the interesting question. Clearly there is SOME information there, the question is whether it is repeatable and linear enough to be useful.

Anyway my little hexapod is now wending its way to me from Illinois (via UPS, not autonomously I'm afraid), Trossen just knocked $100 off the price of a MSR-H01 and that was enough to push me over the edge. My plan is to first build it "stock" with Matt's controller, then start modifying it. I already have some FSRs and an Openservo kit, all I need now is time. The "proprioception" approach is mechanically way easier than using FRSs, so it will be nice if it works.

[ekopramonoid]

Anyone have a GEOMETRY DESIGN from hexapod (MSR-H01) robot,..?
i want to try to build ones,....

best regards

[RayRenteria]

Matt,

I hope you're doing well.

I'm not sure if I'm posting this in the right place so my apologies in advance if this belongs somewhere else.

I can't access the MSR-H01 products online. Please tell me they're still for sale. (http://www.hexapodrobot.com/store/index.php?cPath=21_22) I've read some of the other posts that refer to Trossen but I didn't see it there either. Do you have resellers anywhere that have the kit available?

--Ray