The gravity compensation algorithm fails during individual operation. #83
Description
Describe the problem you got
When I ran the gravity compensation algorithm independently, I found that the values for feedforward compensation were extremely high. This made the mechanical arm prone to curling. Even after scaling down the compensation values overall, the results were not satisfactory. Particularly for the last three joints (567), I feel that they are completely independent of position and often maintain a fixed value. Additionally, the compensation value for joint 5 frequently approaches zero, which is abnormal. I am eagerly looking forward to seeing the visual results shared by someone who successfully ran the gravity compensation algorithm independently.