PiggyCh / RL_arm_under_sparse_reward

A reinforcement learning project for robotic arm under sparse reward
84 stars 11 forks source link

Confusion about .npz files #6

Open dbczhg opened 5 months ago

dbczhg commented 5 months ago

Hello big guy, I'm a newcomer, please advise There are 5 arrays in 'bmirobot_1000_pick_demo.npz': acs: (1000, 100, 4) obs: (1000, 101, 27) info: (1000, 100) g: (1000, 100, 3) ag: (1000, 101, 3) ACS means 1000 pieces of training data for 100 time steps, what does 4 mean? OBS means that there are 1,000 observations, and the initial state is 101 time steps, right? What does 27 represent? info seems to be all {'is_success': 0.0}, probably negligible? Do the '3' of g and ag both represent xyz 3D position information?