search

ryanjulian / embed2learn

Embedding to Learn
9 stars 5 forks source link

MultiTaskEnv with subtasks for Sawyer #52

Closed ryanjulian closed 5 years ago

ryanjulian commented 6 years ago

This is needed for training the embedding with diverse subtasks. The state/action space of all tasks must be merged or the same (tricky).

Initial set of subtasks:

  1. Pick
  2. Place
  3. Pick with obstacles
  4. Place with obstacles
  5. Stack a block on another block
  6. Servo to a location

two possible options:

  1. implement a parameterized version of the SawyerEnvs which switches the reward function based on a parameter(s) and use the existing MultiTaskEnv (which takes a single environment and a dictionary of args/kwargs to initialize it with)
  2. modify MultiTaskEnv to support different environment classes