Autonomous arms

Hi,

I would like to have my arm move independently without user interaction. Which way shall I follow? What kind of hardware do you suggest?

that is just what I want to do :slight_smile:

yes, In the future I plan and entering another science fair with my robot rover and 5DOF arm, and have it autonomouslyrun, so it can then clean my room :smiley:

id suggest one of jims Lynxmotion arms, like the L5 or L6 which can be found here lynxmotion.com/Category.aspx?CategoryID=27

or, you can build you own arm from SES brackets which is also in that same URL under “Custom Aluminum Arms”

they are good platforms to work with, as for the sesors you should have, i would suggest this lynxmotion.com/Product.aspx?productID=260&CategoryID=8

i’ve heard its a good sensor and its easy to interface into a Basic Atom, and it will allow you to tell how far objects are, or you can always get a camera that comes up on your computer screen so you know wat it happening, and theree is a software that allows youto track objects, but thats is more Mikes field of thing (SN96)

dont know how accurate this is, but i hope it helps

I’d recommend a camera.

The CMUcam, CMUcam2, and CMUcam2+ are all nice onboard video processing units that you can interface directly with a microcontroller at a level that the microcontroller can handle.

Or, if you have a PC/laptop handy, you can offload direct video to the computer and have it do all that fancy processing.
If you go that route, a USB webcam would be the cheapes way, or, you could get more fancy with a wireless connection.
I’d recommend Roborealm’s great (FREE!!!) software.

Also, check out the pressure sensors from PhidgetUSA.com
They’d be great to mount on a gripper so that your bot could pick up delicate things, shake hands, etc.

As for controlling the arm, a PC is an easy and powerful choice.
Microcontrollers and onboard processors give it the autonomous feel, though.
Whatever the case, I’d suggest using IK.

o yea, if its autonomous, then i might suggest pressure sensors on the inside of the gripper, so it knows when something is in its grip

Which pressure sensors would you use?

My Arm should be arriving any day soon!!! I think that autonomous or not , pressure sensors are a good idea.

Is there a way to measure a given servos torque?

Neil

Well… yes, technically.
Is there a practical way to do it that would fit on your bot without greatly hindering it?
Not that I can think of.

You can calculate the current draws of the servo by tapping into the pot, but I doubt that’s accurate enough to do anything with.

As for which one I’d use…
I’d use the one who’s dimensions best fit the gripping arms of my bot.
As far as I know, there’s no noticeable accuracy/resolution differences between large or small pressure sensors, or those shaped differently.

Here are some projects that have been done with the arms.

lynxmotion.com/images/html/proj077.htm
lynxmotion.com/images/html/proj075.htm
lynxmotion.com/images/html/proj072.htm
lynxmotion.com/images/html/proj050.htm
lynxmotion.com/images/html/proj002.htm

Not sure of the proper interpretation of autonomous as it pertains to an arm, but all of the projects above do not require human interaction.

Hello
I am also interested in pressure sensors.

Does anyone have a picture of how a pressure sensor can be attached to an arm or finger using the lynxmotion hardware.
Thanks

If its a flexiforce sensor or something like that, I think you can just glue it on :laughing: If its a sensor with its own PCB then you can use double sided tape. I don’t think theres a “real” way to mount something onto the gripper of the arm with SES.

Just be sure that you don’t use regular superglue.
Superglue (a solvent) ate through my first batch of pressure sensors.
:blush:

I ended up using a Loctite Gel superglue that said it wouldn’t damage plastics.
Make sure you don’t forget to get the connectors for the PhidgetUSA pressure sensors.
The leads are paper-thin, so they don’t fit into a .1" breadboard or perfboard.
And, if you try soldering wire directly to them (as I did with the sole surviving one from the first batch) you’ll end up with melted-off leads.
The plastic that the pins grasp will simply melt around the pins.
I was using a friend’s digital soldering iron that was set really low, too.

Hopefully, my previous ineptness will save you some frustration.
:wink:

Thanks for the replies, this is helping to point me in the right direction.

Right now I am leaning towards the flexi pressure sensors and I did investigate the connectors from phidgets.

I will use the SSC32 from lybxmotion for the robot.

Which board would you guys suggest I use to attach the flexi pressure sensors and would this then connect to the SSC32

Also what would be thhe most # of pressure sensors that I could connect ?

Wayne

Well…
FSRs are just variable resistors.

The basic setup is just a voltage divider, by sticking a resistor of known value in series with the FSR.
The measured voltage drop across the FSR will be proportional to the pressure.
This, you can read with a micro’s ADC.

The SSC-32 has 4 extra ADC inputs and the ABB has 3 (if I remember rightly).
A lot of PIC micros have 10 or more.

I’d suggest using a seperate microcontroller, such as an Atom or a PIC, to name the two popular choices, (rather than the SSC-32’s micro) to read the sensors.
That way, you can program it to actually do something with those values have been read.

How many should you use?
oO
Well, if all you need is to tell the pressure of a single gripper, then you need one.
If you want to be more accurate, you can use two and put one on each gripping digit.
Then you can average the two values together to get a more precise readings.

More than two FSRs for the gripper seems a bit redundant to me, though, in most cases.
The only neat case I can think of is a network of tiny FSRs.
I’ve heard of a robotic hand that uses one that it can use to not only determine pressure, but to tell where/how the object is being gripped and whether or not it’s slipping out of the hand.