Stonehenge 3: A fast, precise robot arm

[rghansen]

As you probably noticed, I also modified the cable guide. The Flexi-cables from mindsensors.com needed a bit more support than the orginal Lego cables.

stonehenge_cable_guide.jpg (87.96 KiB) Viewed 13832 times

[rbnnxt]

Thanks for the extra photos. They make the changes to the chute very clear. As you say the track is very critical, both in getting the spacing between the rails right and the positioning of the bottom stop, so that the next ball ends in precisely the right position to be grabbed. I look forward to trying out your mods when I get back to the model next week.

[rghansen]

I couldn't resist trying to make my implementation of the Stonehenge robot arm even faster.

For those interested, the results can be seen below:

[i-fixed-it-44]

Sorry I got to this so late but great it's improved much!

[philoo]

Hi Ron,
I finally rebuilt Stonehenge, adding most of the improvements shown in your previous photos. It does improve reliability of my NXT-G code (I just had to change an angle constant in gripper code). The only problem I had is that the flexicables get entangled in 7-beam you added to the arm (probably my home-made flexicables are not the same length as yours), I changed it to a 5-beam.
But when I tried to run you NXC code, the robot arm initializes properly but the program stops with arm above ball chute lower end. I failed to compile your log decoder, here attached is the logfile. Hopefully you can have a look... Thanks in advance!

[rghansen]

Hi Philo,

The following is your logfile:

0.000: LoopCnt 1
....... sortState: 100
0.001: NumBallsFound 0
....... iStation: 1
0.002: Turn value 0
0.403: Turn value -64
0.404: Rotate error -1
0.579: Open: -42
1.179: Fall: -70
....... Error 200

The arm rotated to the bottom of the chute (station 1) with an error of -1 degree.
It then opened up the gripper.
It then started to drop down, but stopped short of the chute (fall -70).
This caused an Error 200 abort.

What I assume is happening is that the little rubber finger extension on the back finger of the gripper is hitting the guide pin and preventing the arm from descending all the way. If you look at it when it stops, you should see the rubber finger sitting on or close to the guide pin. If you pull the rubber part just 2 millimeters away from the plastic part of the ring finger, then there should be enough clearance between the finger and the guide pin.
Please let me know if this fixes the problem and if you can come up with a more elegant solution to this clearance problem.

[philoo]

Thanks, Ron... Unfortunately it doesn't seem to be the problem. Here is a short movie showing the initialization sequence. The arm stops well above the chute, and nothing blocks it as the slight finger push (after program stop) shows.

[rghansen]

Could it be that the gripper and lifter motor cables have been swapped? It should open up the gripper before lowering the arm, but in the video it doesn't start opening the gripper until after the arm stops descending.

[philoo]

Could it be that the gripper and lifter motor cables have been swapped? It should open up the gripper before lowering the arm, but in the video it doesn't start opening the gripper until after the arm stops descending.

Ah - I'm stupid, I should have had a closer look at the source - indeed the motors were swapped, you didn't use the same order as my version (color sensor port is different too, 2 instead of 1). Works fine now! (just a little problem with the ball that sometimes doesn't remain in the gripper, I'll have a closer look). Thanks!!!

[rghansen]

Actually I should have warned you that I swapped some of the cables around, after all, it's your design. But I'd already forgotten that detail.

In order to get a good grip on the balls, the arm has to descend all the way. If the back finger hits or even rubs on the alignment pin, then the arm will stop descending too soon.