Hello,have you solved the problem yet?

@hantao-zhou @lin711 Is it possible that the method 'vec2quat' is a typo/old name for the method 'pointat2quat' ?
The docstring for this method says : """calculate quaternion from pointat vector"""

@hantao-zhou @lin711 Is it possible that the method 'vec2quat' is a typo/old name for the method 'pointat2quat' ? The docstring for this method says : """calculate quaternion from pointat vector"""

hello！Thank you for your advice! did you try? Will it work after the change？

hello！Thank you for your advice! did you try? Will it work after the change？

I got this error only once on a simulation environment. I was unable to reproduce it. If it comes up again will investigate more. If you can, please do see if this works out and let us know here.

hello！Thank you for your advice! did you try? Will it work after the change？

I got this error only once on a simulation environment. I was unable to reproduce it. If it comes up again will investigate more. If you can, please do see if this works out and let us know here.

I was wondering whether vec2quat is related to the grasping posture of the robotic arm. Although I changed vec2quat to pointat2quat according to your suggestion, I found that I gave the command to grasp the cup from its side but in the simulation the robotic arm still grabs the cup from top to bottom, Do you know how to grasp the cup vertically from its side (that is, change the grasp attitude of the robot arm)? I would really appreciate any reply from you！Thanks

This task of grasping the cup. Which simulation task is this? Is it a part of RLBench? I haven't seen this till now.

This task of grasping the cup. Which simulation task is this? Is it a part of RLBench? I haven't seen this till now.

the task is pick_up_cup which is in Voxposer /RLBench/rlbench/tasks. Also, may I ask if you have considered migrating the real machine? I am very confused about the labeling issue of objects after migrating the real machine. Thank you very much for your reply~~~~

Hello, I carefully studied the relevant part of the code and found a serious problem. The rotation axis calculated in the 'pointat2quat' function for the quaternion does not match the rotation axis of the end effector's quaternion. This inconsistency is likely the reason for the errors in execution. I personally cannot understand the choice of a random vector in 'pointat2quat.'I suggest using the direction vector of the end effector's rotation axis as the direction vector for 'pointat2quat.'
你好，我仔细研究了相关代码部分，发现里面有一个致命的问题。计算四元数的‘pointat2quat’部分的旋转轴，和末端执行器四元数的旋转轴不一致。这很有可能是导致执行效果出错的原因。个人无法理解‘pointat2quat’中取一个随机向量的行为。建议将末端执行器旋转轴的方向向量作为‘pointat2quat’的方向向量。