File tree Expand file tree Collapse file tree
Expand file tree Collapse file tree Original file line number Diff line number Diff line change @@ -119,10 +119,10 @@ Observation Space
119119Rewards
120120-------
121121
122- **Sparse **: ``0.0 `` if ``‖ee − goal‖ < 0.05 `` else ``-1.0 ``.
122+ **Sparse **: ``0.0 `` if ``‖ee − goal‖ < 0.02 `` else ``-1.0 ``.
123123
124124**Dense **: same shape as the other robots' reach envs. Defaults
125- from ``config/ned2_reach_task_config.yaml ``: ``reach_tolerance=0.05 ``,
125+ from ``config/ned2_reach_task_config.yaml ``: ``reach_tolerance=0.02 ``,
126126``multiplier_dist_reward=2.0 ``, ``reached_goal_reward=20 ``,
127127``step_reward=-0.5 ``.
128128
Original file line number Diff line number Diff line change @@ -118,11 +118,11 @@ z = 0.78.)
118118Rewards
119119-------
120120
121- **Sparse **: ``0.0 `` if ``‖ee − goal‖ < 0.05 `` else ``-1.0 ``.
121+ **Sparse **: ``0.0 `` if ``‖ee − goal‖ < 0.02 `` else ``-1.0 ``.
122122
123123**Dense **: dist-shaped + reached-goal bonus + per-step penalty +
124124joint/none/goal-space penalties. Defaults from
125- ``config/rx200_reach_task_config.yaml ``: ``reach_tolerance=0.05 ``,
125+ ``config/rx200_reach_task_config.yaml ``: ``reach_tolerance=0.02 ``,
126126``multiplier_dist_reward=2.0 ``, ``reached_goal_reward=20 ``,
127127``step_reward=-0.5 ``, ``joint_limits_reward=-2.0 ``,
128128``none_exe_reward=-5.0 ``, ``not_within_goal_space_reward=-2.0 ``.
Original file line number Diff line number Diff line change @@ -232,7 +232,7 @@ kwarg.
232232 + not_within_goal_space_reward if goal sampling failed
233233
234234 Defaults (from ``config/ur5e_reach_task_config.yaml ``):
235- ``reach_tolerance=0.05 ``, ``multiplier_dist_reward=2.0 ``,
235+ ``reach_tolerance=0.02 ``, ``multiplier_dist_reward=2.0 ``,
236236``reached_goal_reward=20 ``, ``step_reward=-0.5 ``,
237237``joint_limits_reward=-2.0 ``, ``none_exe_reward=-5.0 ``,
238238``not_within_goal_space_reward=-2.0 ``.
Original file line number Diff line number Diff line change @@ -172,13 +172,13 @@ world z = 0.78 — so a goal z of 0.20 is 1.0 m above the floor.)
172172Rewards
173173-------
174174
175- **Sparse ** (required for HER): ``0.0 `` if ``‖ee − goal‖ < 0.05 `` else
175+ **Sparse ** (required for HER): ``0.0 `` if ``‖ee − goal‖ < 0.02 `` else
176176``-1.0 ``.
177177
178178**Dense ** (default for std env): dist-shaped penalty + reached-goal
179179bonus + per-step penalty + joint-limit / non-executable / not-in-goal-
180180space penalties. Defaults from ``config/vx300s_reach_task_config.yaml ``:
181- ``reach_tolerance=0.05 ``, ``multiplier_dist_reward=2.0 ``,
181+ ``reach_tolerance=0.02 ``, ``multiplier_dist_reward=2.0 ``,
182182``reached_goal_reward=20 ``, ``step_reward=-0.5 ``,
183183``joint_limits_reward=-2.0 ``, ``none_exe_reward=-5.0 ``,
184184``not_within_goal_space_reward=-2.0 ``.
You can’t perform that action at this time.
0 commit comments