Trifinger - Homing failure causing motors to run violently

Hi everyone,

I’ve recently run some tests on the Trifinger robot and got some weird issues that caused one of the fingers broken.

Before running all three fingers, I separately tested each finger and there was no problem. I run script demo_single_finger_position_control.py in robot_fingers package and used next_index as homing method. Everything went smoothly without error, so I’m confident that there is no problem with the hardware, and also PD gains are good.

However, when I attempted to run script demo_trifingeredu.py, also with homing method next_index, I noticed that all three fingers did not slowly find index ticks on encoders as usual, they just did not move. Then suddenly, some joints started shaking and moved violently even though the PD gains were set to pretty small values. I managed to cut off the power but it was too late, one finger hit the frame and broke.

Has anyone experienced similar issue? I suspect that the problem might come from CAN communication, maybe delays or something. When only one finger is running, there are only two active CAN channels, but we have 6 channels when all three are running.

Thanks!

Hello,

I also observe this (i.e. some joint moving violently back and forth) sometimes when running something the first time after powering the robot on. Sorry to hear that it broke a finger in your case!
Unfortunately I have no idea what is causing this. It happens only very rarely for me, so I don’t see a good way of debugging it. I simply got used to staying close to the emergency stop when I run something the first time after powering on. I never saw this happening on consecutive runs while the power is kept on, so it might be some issue during initialisation in the motor board firmware.

Best,
Felix

Hello,
I also observed this (i.e. some joints move violently back and forth). My solution is to restart the motor driver board (μ Driver) by pressing the button after power on.
Have you found the cause of this malfunction?