Skip to content

Latest commit

 

History

History
42 lines (22 loc) · 1.05 KB

File metadata and controls

42 lines (22 loc) · 1.05 KB

ForceRL

Depandencies

Some depandencies are in environment.yml file

Fixing issue for rendering

type this in terminal before running code to get around OpenGL issue

export MUJOCO_GL="osmesa"

To permenantly get rid of this issue, add the above line to your .bashrc file.

Environment TODOs

original_door_env.py contains the original door environment copied from robosuite. It contains just a door without robot.

curri_door_env.py contains the door environment for curricula of door pose initialized in increasing range. It extends original door env.

door_random_point_env.py contains the door environment that generates random force point each episodes. It extends curriclum door env.

Findings

Curriculum learning with

fixed initial pose -> wider initial pose -> random point

works good.

Do we need discretization if so?

Revolute object Axis:

For microwave, X is facing outward, Y is facing up, Z is facing "left" dishwasher: X facing outward, Y facing downward, Z facing "right" Door counterclock: X out, Y left, Z down