changelog.md

Change log

• 🚀 for RofuncRL updates
• 🚄 for planning and control updates
• 🖼️ for Visualab updates
• 📈 for Datalab updates
• 📚 for documentation updates
• 🎮 for simulator updates
• 🐛 for bug fixes
• 🏄‍♂️ for system updates
• 🎉 for event celebrations

Update News 🎉🎉🎉

• [2024-01-24] 🚀 CURI Synergy-based Softhand grasping tasks are supported to be trained by RofuncRL.
• [2023-12-24] 🚀 Dexterous hand (Shadow Hand, Allegro Hand, qbSofthand) tasks are supported to be trained by RofuncRL.
• [2023-12-07] 🖼️ EfficientSAM is supported for high-speed segmentation on edge devices like Nvidia Jetson, check the example in Visualab.
• [2023-12-04] 🖼️ VLPart-SAM is supported for part-level segmentation with text prompt, check the example.
• [2023-12-03] 🖼️ Segment-Anything (SAM) is supported in an interactive mode, check the examples in Visualab (segment anything, segment with prompt).
• [2023-10-31] 🚀 RofuncRL: A modular easy-to-use Reinforcement Learning sub-package designed for Robot Learning tasks is released. It has been tested with simulators like OpenAIGym, IsaacGym, OmniIsaacGym (see example gallery), and also differentiable simulators like PlasticineLab and DiffCloth.
• [2023-10-31] 🚀 RofuncRL supports OmniIsaacGym tasks.

v0.0.2.5 More examples

• [2023-10-16] 🎉 More examples are added for RofuncRL and planning and control modules.
• [2023-07-12] 🎉 BiRP paper is accepted by IEEE CDC 2023. Check the arxiv version here.
• [2023-06-24] 🎉 SoftGPT paper is accepted by IROS2023. Check the arxiv version here.

v0.0.2

• [2023-06-13] 🎉 This is the second official release version of Rofunc package. We provide a reinforcement learning baseline framework (RofuncRL) that performs well in robot tasks 🦾, and several state-of-the-art online reinforcement learning algorithms (PPO, SAC, and TD3) have been completed.🥳
• [2023-05-21] 🚄 LQT (Linear Quadratic Tracking), iLQR (iterative Linear Quadratic Regulator) and their variants are supported in planning and control module. Check the example gallery.

v0.0.1

• [2022-12-17] 🎉 This is the first offlical release version Rofunc package. The four core parts (Multimodal sensors, RL baselines, Control and Isaac simulator) are initially supported. 🎉

  • Devices: support Xsens, optitrack, Zed 2i (multi), Delsys EMG, Manus gloves and multimodal
  • Learning algorithms: TP-GMM, TP-GMR, RL baselines (RLlib, ElegantRL, SKRL with IsaacGym)
  • Planning and control: LQT variants, iLQR variants
  • Simulator (IsaacGym-based): CURI, Ubtech walker, Franka, Baxter, Sawyer, CURI-mini