wangzixiang
/
lerobot
mirror of https://github.com/huggingface/lerobot.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Add more information about Slate Base

This commit is contained in:
shantanu-tr 2025-04-03 15:39:21 -05:00
parent 033f3cb880
commit 0844e0e365
1 changed files with 22 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
examples/13_use_trossen_ai_mobile.md
View File

@ -60,11 +60,11 @@ python lerobot/scripts/control_robot.py \
--robot.max_relative_target=null \
--control.type=teleoperate
```
By adding `--robot.force_feedback_gain=0.1`, we override the default value for `force_feedback_gain` defined in [`TrossenAIBimanualRobot`](lerobot/common/robot_devices/robots/configs.py). This enables **force feedback** from the follower arm to the leader arm — meaning the user can **feel contact forces** when the robot interacts with external objects (e.g., gripping or bumping into something). A typical starting value is `0.1` for a responsive feel. You can disable this behavior entirely by setting `--robot.force_feedback_gain=0.0` in the command line:
By adding `--robot.force_feedback_gain=0.1`, we override the default value for `force_feedback_gain` defined in [`TrossenAIMobileRobot`](lerobot/common/robot_devices/robots/configs.py). This enables **force feedback** from the follower arm to the leader arm — meaning the user can **feel contact forces** when the robot interacts with external objects (e.g., gripping or bumping into something). A typical starting value is `0.1` for a responsive feel. You can disable this behavior entirely by setting `--robot.force_feedback_gain=0.0` in the command line:
```bash
python lerobot/scripts/control_robot.py \
--robot.type=trossen_ai_stationary \
--robot.type=trossen_ai_mobile \
--robot.max_relative_target=null \
--robot.force_feedback_gain=0.1 \
--control.type=teleoperate
@ -110,6 +110,19 @@ python lerobot/scripts/control_robot.py \
--control.display_cameras=false
```
The **Slate base** works in two modes:
- **Torque OFF** (default): You can push the base around manually.
- **Torque ON**: Enables the motors so you can control the base using the **Slate remote controller**.
To enable torque-on mode during recording, add the following argument:
```bash
--robot.enable_motor_torque=true
```
For more information about the Slate remote controller, refer to the official documentation:
[Slate RC Controller Guide](https://docs.trossenrobotics.com/slate_docs/operation/rc_controller.html)
## Visualize a dataset
If you uploaded your dataset to the hub with `--control.push_to_hub=true`, you can [visualize your dataset online](https://huggingface.co/spaces/lerobot/visualize_dataset) by copy pasting your repo id given by:
@ -138,6 +151,8 @@ python lerobot/scripts/control_robot.py \
--robot.enable_motor_torque=true
```
Note: For replaying an episode, you need to turn on motor torque using ``--robot.enable_motor_torque=true``, so that the robot can actively follow the trajectory instead of remaining in a passive (torque-off) state.
## Train a policy
To train a policy to control your robot, use the [`python lerobot/scripts/train.py`](../lerobot/scripts/train.py) script. A few arguments are required. Here is an example command:
@ -178,9 +193,13 @@ python lerobot/scripts/control_robot.py \
--control.num_episodes=10 \
--control.push_to_hub=true \
--control.policy.path=outputs/train/act_trossen_ai_mobile_test/checkpoints/last/pretrained_model \
--control.num_image_writer_processes=1
--control.num_image_writer_processes=1 \
--robot.enable_motor_torque=true
```
Note: For evaluation, you need to turn on motor torque using ``--robot.enable_motor_torque=true``, so that the robot can actively follow the trajectory instead of remaining in a passive (torque-off) state.
As you can see, it's almost the same command as previously used to record your training dataset. Two things changed:
1. There is an additional `--control.policy.path` argument which indicates the path to your policy checkpoint with (e.g. `outputs/train/eval_act_trossen_ai_mobile_test/checkpoints/last/pretrained_model`). You can also use the model repository if you uploaded a model checkpoint to the hub (e.g. `${HF_USER}/act_trossen_ai_mobile_test`).
2. The name of dataset begins by `eval` to reflect that you are running inference (e.g. `${HF_USER}/eval_act_trossen_ai_mobile_test`).