GibsonEnv icon indicating copy to clipboard operation
GibsonEnv copied to clipboard

Published 20 hours ago • verified publisher icon StanfordVL

Target Orientation Issue

Open Berk035 opened this issue 5 years ago • 1 comments

Hello everyone,

I studied navigation on husky with 'pposgd_fuse' and only 'depth-sensor' values. The agent could able to reach target point. After, it reached target point total reward began decrease due to orientation couldn't obtain. How can I handle this problem without removing 'angle_cost' from rewards array? Thanks in advance.

Berk035 avatar Jan 23 '20 07:01 Berk035

Hi again,

In my opinion, after the agent reached to target point, the different process must be applied. Sincerely.

Berk035 avatar Feb 14 '20 12:02 Berk035