Robotics
LeRobot
English
OpenRAL
rskill
diffusion
vision-language-action
pusht
diffusion-policy
manipulation
Instructions to use OpenRAL/rskill-diffusion-pusht-pusht-fp32 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LeRobot
How to use OpenRAL/rskill-diffusion-pusht-pusht-fp32 with LeRobot:
- Notebooks
- Google Colab
- Kaggle
rskill.yaml: declare action_contract.joint_units: radians (issue #135)
Browse files- rskill.yaml +5 -0
rskill.yaml
CHANGED
|
@@ -96,3 +96,8 @@ scenes:
|
|
| 96 |
# to bind the LeRobot v3 `action` feature shape).
|
| 97 |
action_contract:
|
| 98 |
dim: 2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 96 |
# to bind the LeRobot v3 `action` feature shape).
|
| 97 |
action_contract:
|
| 98 |
dim: 2
|
| 99 |
+
# issue #135 — EXPLICIT joint units, verified against the checkpoint normalizer
|
| 100 |
+
# stats. The skill_runner converts deg<->rad at the policy boundary; declaring
|
| 101 |
+
# it here replaces the fragile stats-magnitude guess (which drove a real arm
|
| 102 |
+
# into its limits). RADIANS = no conversion.
|
| 103 |
+
joint_units: "radians"
|