To eliminate the slop problem I used some elastic bands to pre-load the servos. You're not going to need much but you can set them to work against gravity on the arm (think springs on an angle poise lamp). Do you think an elastic band around the gripper would help?
When doing the "smooth" program remember, servos get data at 50Hz so you can not update faster than that. I made the mistake of trying to loop faster and found I was restarting the PCA9685 before it had sent a full pulse to the servo. Bad move, bad bad move, servo hit end stop, go boom.
As part of the "smooth" routine you write an "up" routine that brings the gripper directly up/down an axis rather than a rotation around the two arm servos. That should give you something to bend your mind around. ?
The 600 series had rubber skin. We spotted them easy, but these are new. They look human... sweat, bad breath, everything. Very hard to spot.
The only important slop error I'm currently having is in the base rotation. When the arm is fully extended the position of the claw can be up to a full inch off target right or left. And that will represent the maximum error. I haven't really noticed any up and down slop in the arm. That much seems to be pretty consistent.
And yes, the problem of being able to lift things straight up and lower them straight down is precisely the problem I'm hoping to work on. In fact, that's the problem I'm currently working on when the side-to-side error of the base became a problem.
I'm not going to give up on it just yet. I only just worked on this today. But I don't want to spend all week playing around with this Me Arm. I have things to do that are actually important. ?
It's 2:00 am and I'll still playing around with this stupid Me Arm. I gotta quit.
Thanks for the suggestions, you've been a great help. I'll go do some other things tomorrow and come back to this Me Arm at some later date.
RoboPi wrote: And yes, the problem of being able to lift things straight up and lower them straight down is precisely the problem I'm hoping to work on.
The change in angles required to move along any straight (or curved) path can be computed in real time (no tables required). The interaction between motors you mention could also be modelled although I would simply make or get a better robotic arm to play with. This is illustrated in the program I posted that returns the angles for any mx,my position. Instead or using the mouse you could feed it positions from a path generator program and it will output the desired angles which is what I thought you wanted. The output below is from an actual program I have not used PAINT this time 🙂
RoboPi wrote: And yes, the problem of being able to lift things straight up and lower them straight down is precisely the problem I'm hoping to work on.
The change in angles required to move along any straight (or curved) path can be computed in real time (no tables required). The interaction between motors you mention could also be modelled although I would simply make or get a better robotic arm to play with. This is illustrated in the program I posted that returns the angles for any mx,my position. Instead or using the mouse you could feed it positions from a path generator program and it will output the desired angles which is what I thought you wanted. The output below is from an actual program I have not used PAINT this time 🙂
The problem is that the servo angles on the Me Arm do not represent the actual angles of the arm members. Although their may be a linear relationship between the servo angle and the upper arm of the Me Arm. However, the angle of the lower (or outer) arm of the Me Arm appears to me to have an even more complex relationship with the servo angle that drives it. This is because the actual angle of the outer arm can depend on the position of the upper arm.
None the less there may be ways to create a map between the actual arm angles and the angles of their servo motors. I'm not at that stage yet. I'm still trying to get the Me Arm to behave itself. ?
I have coordinates that I would like to have it run through and I'm currently in the process of programming the Mega to read the position data from the SD card and turn that into actual arm motion. I'm getting there. But not quite finished yet.
So here is an output generated by the program for the angles required to move an arm down along the Y axis with the X axis fixed at 150.
Yes. And there is also no point in all the decimal places since the servos can only accept integer steps in degrees anyway.
In your example, my program will only need the first and last entries on your list saved to the SD card. All the intermediate positions will be calculated programmatically using a "Smooth Motion" method.
In fact, without using "Smooth Motion" here is all the data I need to pick an eyedropper straight up out of the test tube rack:
090,000,045,000 <- Home Position 005,070,070,055 <- moves arm to the eyedropper location & opens claw 005,080,090,055 <- moves the arm toward the eyedropper. 005,080,090,000 <- closes the claw 005,075,095,000 <- from here down is the straight up lift. 005,070,100,000 005,065,110,000 005,060,120,000 005,055,130,000 005,050,160,000 <- the eyedropper is now clear of the rack.
This is my first set of motion coordinates. It works manually, now I'm working on having it done via reading these inputs from the SD card. I'm working on this software routine as we speak. So hopefully I'll be able to report back with some success here shortly.
I finally got the SD card reading method to work. Mandy had warned me about using too quick of delays and that was my problem. I made the delays between each servo update a little longer and now it's working. Unfortunately it's still not able to pick up even the eyedropper. The motion of the arm is too jerky without the "Smooth Motion" routine and it slips right off the eyedropper bulb. I replaced the eyedropper with a cardboard nail file and it lifted the nail file up with no problem.
I'm going to try finding a more powerful power supply as Mandy had suggested to see if I can give the gripper a bit more holding power. Other than this the arm went right over to the proper place, and picked up the cardboard nail file.
So I'm making progress.
I need to quit working on this for now as I have real life things falling too far behind! ?
But this is a good place to stop. At least I have the arm calibrated and up to the point where it can follow a path described by a data set on the SD card. That's a far cry further along than I was last week. ?
If the claw is trying to fully closed (get to position 000) but the test tube is in the way the servo will continue to try to close pulling power resulting in a possible overload of the circuit or the claw flexing under load. This may be why it finds it easy to pick up a thin object rather than the thicker.
The claw should stop trying to close once it has a grip on the object.
As the test tube is wider at the top you only need to have the claw closed enough to stop the top falling through.
My suggestion would be to fit a MAX471 current sensor on the 5V to the servos (V+), it gives an output of 1V per 1Amp. Measure the output with an analogue pin and this way you can monitor the current to the servos and stop movement if they pull too much power. You will be able to monitor the grip on the claw this way.
I'm not sure about your co-ordinates though, I'm assuming 0 to 55 degrees for the jaw movement, that is not a lot of options for positioning. Accessing the PCA9685 directly I get steps from 250 to 1300 for 180 degrees of travel or 320 steps over 55 degrees, nearly six times the resolution.
Does this help?
This post was modified 5 years ago 2 times by Mandy
The 600 series had rubber skin. We spotted them easy, but these are new. They look human... sweat, bad breath, everything. Very hard to spot.
My suggestion would be to fit a MAX471 current sensor on the 5V to the servos (V+), it gives an output of 1V per 1Amp. Measure the output with an analogue pin and this way you can monitor the current to the servos and stop movement if they pull too much power. You will be able to monitor the grip on the claw this way.
That definitely sounds interesting. I'll have to look into that. I don't have a MAX471 current sensor here but I might be able to rig something up to measure the current using an analog pin.
In the meantime I put a cork in it.
I put a small cork in the test tube and the Me Arm was able to lift the test tube by the cork. ?
I'm also thinking that I might be able to find some kind of foam or soft rubber around here that I can slip onto the claw jaws. That will give the servo something to squeeze on. That should give it a little more of a springy action.
It lifts the test tube out of the rack. So far so good. But now to get it to move the test tube over to another hole and put it back into the rack will be an even bigger challenge because that will need to be done with even more accuracy.
This is so much fun! But my real life is taking a beating in terms of me not getting my chores done. It's hot and muggy outside so I'm using that as an excuse. Plus it's the 4th of July weekend. Gotta be allowed to play with my robot stuff over the 4th right?
I finally wrote up a "Smooth Motion" routine to have the program control the Me Arm to lift "straight up". I put "straight-up" in quotes because it's not prefect, but it's pretty darn close. In fact, I don't think the Me Arm could be asked to do much better.
In any case, what is needed is the starting position and the ending position. So I do need to have those coordinates known. Although it may be possible later to calculate what a final position needs to be.
In any case. My data set has changed and the SD card now contains the following data:
This is actually less data than previously. But I have also added a leading letter to indicate the action required. "M" means to just move to that location directly. No need for smooth motion when just moving from a home position over to the test tube rack. Just let her rip. So I start off with three move commands.
The commands that begin with "L" signal a lift command that needs to be lifted straight up. So when the program see the "L" it calls the smooth motion lifting routine. Notice that the last two entries in the above table differ by quite a bit in the middle two coordinates. But the outer coordinates are constant. This is because when lifting straight up these are the only two numbers that need to change.
So what I needed to do next was to map 100 to 075 with 090 to 160. And break both of these up into equal parts to generate a bunch of intermediate coordinates. I do this by calculating the difference of one set and then using that difference to map both coordinates to. This way they should step along in close proximity with each other. It worked.
The actual program is quite complicated so I'll just post a serial output routine here to show what's going on.
/*
* Generic Input Routines
* by Robot Pi
* Variables are defined within the methods
*/
void setup() {
Serial.begin(9600); // Use Serial monitor :
}
void loop() {
mapping_Test();
}
void mapping_Test()
{
String str_input;
int num_1; int num_2; int num_3; int num_4; // define vars
Serial.println("Mapping Test");
Serial.println("Input four 3-digit integers, separated by commas.");
while (Serial.available() == 0){} // wait for input
str_input = Serial.readString();
// How to count for substring eg. 012,456,890,234 = 012(3)456(7)890(11)234
// Also converting String.toInt() while here.
num_1 = str_input.substring(0,3).toInt();
num_2 = str_input.substring(4,7).toInt();
num_3 = str_input.substring(8,11).toInt();
num_4 = str_input.substring(12).toInt();
// Print results to Serial Monitor
Serial.print("num_1 = "); Serial.println(num_1,DEC);
Serial.print("num_2 = "); Serial.println(num_2,DEC);
Serial.print("num_3 = "); Serial.println(num_3,DEC);
Serial.print("num_4 = "); Serial.println(num_4,DEC);
int diff; // Difference value
int mapped_answer;
diff = abs(num_3-num_4);// Find absolute difference between last two numbers
// Map out over the difference of the last two numbers.
for (int map_out = 1; map_out < diff; map_out++)
{
mapped_answer = map(map_out, 0, diff, num_1, num_2); // map the first two numbers
Serial.print(mapped_answer,DEC); Serial.print(" : ");
mapped_answer = map(map_out, 0, diff, num_3, num_4); // map the second two numbers
Serial.println(mapped_answer,DEC);
}
}
Putting in the numbers I mentioned above produces the following output:
That would be a lot of data to have to type into a text file. Not to mention that I wouldn't even know the proper mapping anyway. So this way the program generates these coordinates using this mapping technique. And sends them to the Me Arm one after the other to control the Me Arm to lift "straight up".
Like as I say, it's not perfect, but it's pretty darn close and it does work from a practical perspective. I also don't believe the Me Arm could do much better in any case. It's just not that precise.
So now, as you can see, with just a very limited amount of coordinates on the SD text file I can generate the missing intermediate coordinates to achieve "Smooth Motion".
As I say, the actual program that does this is quite a bit more complicated but I'll be more than happy to post the code if anyone would like to see it.
I think I'm going to stop at this point and look into creating my own Arduino Libraries so I can place the SD card reading methods and the Smooth Motion method into libraries and clean up the main code a bit. My 70 year old eyes are starting to bug out at this point. I need to clean things up a bit. ?
We use cookies on the DroneBot Workshop Forums to give you the most relevant experience by remembering your preferences and repeat visits. By clicking “Accept”, you consent to the use of ALL the cookies.
This website uses cookies to improve your experience while you navigate through the website. Out of these, the cookies that are categorized as necessary are stored on your browser as they are essential for the working of basic functionalities of the website. We also use third-party cookies that help us analyze and understand how you use this website. These cookies will be stored in your browser only with your consent. You also have the option to opt-out of these cookies. But opting out of some of these cookies may affect your browsing experience.
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
Any cookies that may not be particularly necessary for the website to function and is used specifically to collect user personal data via analytics, ads, other embedded contents are termed as non-necessary cookies. It is mandatory to procure user consent prior to running these cookies on your website.
