ROS Index
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |
Launch files
Messages
Services
Plugins
Recent questions tagged furuta_pendulum_rl at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |
Launch files
Messages
Services
Plugins
Recent questions tagged furuta_pendulum_rl at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |
Launch files
Messages
Services
Plugins
Recent questions tagged furuta_pendulum_rl at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |
Launch files
Messages
Services
Plugins
Recent questions tagged furuta_pendulum_rl at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |
Launch files
Messages
Services
Plugins
Recent questions tagged furuta_pendulum_rl at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |
Launch files
Messages
Services
Plugins
Recent questions tagged furuta_pendulum_rl at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |
Launch files
Messages
Services
Plugins
Recent questions tagged furuta_pendulum_rl at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |
Launch files
Messages
Services
Plugins
Recent questions tagged furuta_pendulum_rl at Robotics Stack Exchange
Package Summary
| Tags | No category tags. |
| Version | 1.0.0 |
| License | GPLv3 |
| Build type | AMENT_CMAKE |
| Use | RECOMMENDED |
Repository Summary
| Description | LQR, MPC and DRL approaches to control the Furuta pendulum. |
| Checkout URI | https://github.com/macstepien/furuta_pendulum.git |
| VCS Type | git |
| VCS Version | master |
| Last Updated | 2024-04-14 |
| Dev Status | UNKNOWN |
| Released | UNRELEASED |
| Tags | control robotics ros inverted-pendulum |
| Contributing |
Help Wanted (-)
Good First Issues (-) Pull Requests to Review (-) |
Package Description
Additional Links
Maintainers
- Maciej Stępień
Authors
- Maciej Stępień
furuta_pendulum_rl
Setup
Mujoco
Copy mujoco210 to ~/.mujoco/mujoco210
Imitation
imitation library isn’t currently installed automatically, so it is necessary to do it manually:
pip install imitation
[!WARNING]
It may result in upgrading gym, and it will be necessary to downgrade it again (by once again executing install of the specific stable baselines version, refer to Dockerfile). Here is the message caused by this problem:ImportError: Missing shimmy installation. You provided an OpenAI Gym environment. Stable-Baselines3 (SB3) has transitioned to using Gymnasium internally. In order to use OpenAI Gym environments with SB3, you need to install shimmy (`pip install 'shimmy>=0.2.1'`).
Training
To solve Furuta control, I created three environments with progressive level of difficulty:
- furuta_pendulum_upward_stabilization - pendulum starts with some angle that is a little off upward position and with some velocity, the agent’s task is to stabilize in the 0 position. A reward is received for every episode, in which the angle isn’t greater than some threshold.
- furuta_pendulum_swing_up - pendulum starts in the downward stable position, the agent’s task is to swing it up close to the upward position
- furuta_pendulum_full - this is a full Furuta pendulum control problem - the pendulum starts in the downward stable position, and the agent’s task is to swing it up and stabilize it upward.
Solutions
In scripts with the name of the environment and the name of the algorithm you can find example solutions.
Exporting model
Due to problems with ROS2 performance in Python I decided to use C++. To export the actor model from SAC used to solve the full control problem, I used export_sac_model.py script.
Development
Problems
High variance of actions - in gaussian exploration actions can be different in each timestep which results in high variance and oscillations of the motor. gSDE results in smoother actions.
Changelog for package furuta_pendulum_rl
1.0.0 (2023-07-30)
- Working simulation
- Contributors: Maciej Stępień
Package Dependencies
| Deps | Name |
|---|---|
| sensor_msgs | |
| std_msgs | |
| ament_cmake | |
| launch | |
| launch_ros |
System Dependencies
Dependant Packages
| Name | Deps |
|---|---|
| furuta_pendulum_bringup |