DiffSim2Real: Deploying Quadrupedal Locomotion Policies Purely Trained in Differentiable Simulation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913571244343296 |
|---|---|
| author | Bagajo, Joshua Schwarke, Clemens Klemm, Victor Georgiev, Ignat Sleiman, Jean-Pierre Tordesillas, Jesus Garg, Animesh Hutter, Marco |
| author_facet | Bagajo, Joshua Schwarke, Clemens Klemm, Victor Georgiev, Ignat Sleiman, Jean-Pierre Tordesillas, Jesus Garg, Animesh Hutter, Marco |
| contents | Differentiable simulators provide analytic gradients, enabling more sample-efficient learning algorithms and paving the way for data intensive learning tasks such as learning from images. In this work, we demonstrate that locomotion policies trained with analytic gradients from a differentiable simulator can be successfully transferred to the real world. Typically, simulators that offer informative gradients lack the physical accuracy needed for sim-to-real transfer, and vice-versa. A key factor in our success is a smooth contact model that combines informative gradients with physical accuracy, ensuring effective transfer of learned behaviors. To the best of our knowledge, this is the first time a real quadrupedal robot is able to locomote after training exclusively in a differentiable simulation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_02189 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | DiffSim2Real: Deploying Quadrupedal Locomotion Policies Purely Trained in Differentiable Simulation Bagajo, Joshua Schwarke, Clemens Klemm, Victor Georgiev, Ignat Sleiman, Jean-Pierre Tordesillas, Jesus Garg, Animesh Hutter, Marco Robotics Machine Learning I.2.9; I.6.5 Differentiable simulators provide analytic gradients, enabling more sample-efficient learning algorithms and paving the way for data intensive learning tasks such as learning from images. In this work, we demonstrate that locomotion policies trained with analytic gradients from a differentiable simulator can be successfully transferred to the real world. Typically, simulators that offer informative gradients lack the physical accuracy needed for sim-to-real transfer, and vice-versa. A key factor in our success is a smooth contact model that combines informative gradients with physical accuracy, ensuring effective transfer of learned behaviors. To the best of our knowledge, this is the first time a real quadrupedal robot is able to locomote after training exclusively in a differentiable simulation. |
| title | DiffSim2Real: Deploying Quadrupedal Locomotion Policies Purely Trained in Differentiable Simulation |
| topic | Robotics Machine Learning I.2.9; I.6.5 |
| url | https://arxiv.org/abs/2411.02189 |