search

andyzeng / visual-pushing-grasping

Train robotic agents to learn to plan pushing and grasping actions for manipulation with deep reinforcement learning.
http://vpg.cs.princeton.edu/
BSD 2-Clause "Simplified" License
898 stars 315 forks source link

where could I see the 'completion' in the paper #38

Open cww97 opened 5 years ago

cww97 commented 5 years ago

in you paper(table 1 & table 2), we can see the 'completion', 'grasp success' and 'action efficiency' of each method. however, when I run evaluate.py for a test

python evaluate.py --session_directory 'logs/2019-08-27.12:22:52' --method 'reinforcement' --num_obj_complete 30

I wanna see the third line of your table 1, then I get

Average % clearance: 60.0
Average % grasp success per clearance: 73.5
Average % action efficiency: 67.9
Average grasp to push ratio: 92.0

is the 'clearance' is the 'completion' in your table ?

I see the vpg method get 100 completion. Is where I see the completion wrong ? or the model is not right? I train the reinforcement for 25000 rounds, the frist 2500 round is familer with the fig4 in your table.

QAQ