The award program has been organized to reward the best open source contributions of 2023. All contributions were eligible for application: it could be an open-source plugin, a pull-request on GitHub, a video, a tutorial or documentation pages.
The SOFA Award 2020 strongly motivated me to dig into the code of SOFA and contribute way more to the open-source!
The program included two prizes. The winners have been announced on 14th November during the 2023 SOFA Symposium:
| Organizers’ Prize Prize: 600€ |
Public Prize Prize: 400€ |
|---|---|
| Protopapa, Andrea (Politecnico di Torino, Italy) the SOFA-DR-RL project |
Scheikl, Paul Maria (KIT, Germany) the LapGym project |
| This work demonstrates how Domain Randomization (DR) improves RL policies for soft robots by enhancing robustness and reducing training time. We introduce RF-DROPO for sim-to-real transfer and provide a user-friendly extension of the SofaGym framework for DR-compatible tasks and implementation. This toolkit uses Stable Baselines3 (SB3) for Reinforcement Learning training, making it easier to create SOFA scenes and control policies for Sim2Real transfer with example scenes for guidance. | We present LapGym, a framework for building Reinforcement Learning (RL) environments for RALS that models the challenges posed by surgical tasks, and sofa_env, a diverse suite of 12 environments. Motivated by surgical training, these environments are organized into 4 tracks: Spatial Reasoning, Deformable Object Manipulation & Grasping, Dissection, and Thread Manipulation. Each environment is highly parametrizable for increasing difficulty, resulting in a high performance ceiling for new algorithms. |
Find all award candidates on the 2023 Award page here.
Congratulations to Andrea & Paul!
Comments are closed.