Generating an observation without calling env.step, particularly after resetting physics state

google-deepmind / dm_env

A Python interface for reinforcement learning environments

Apache License 2.0

342 stars 43 forks source link

Generating an observation without calling env.step, particularly after resetting physics state #16

Closed StoneT2000 closed 4 months ago

StoneT2000 commented 1 year ago

Was wondering if I could reset an environment to a particular state and instead of calling env.step to get a timestep and observation, I could call a function to get the current timestep information directly. Is this possible?