robosuite
multi-task-tcc-robosuite
XIRL
Overview
Setup
Datasets
Code Navigation
Experiments: Reproducing Paper Results
Extending XIRL
Acknowledgments
Overview
Code release for our CoRL 2021 conference paper:
XIRL: Cross-embodiment Inverse Reinforcement Learning
Kevin Zakka1,3, Andy Zeng1, Pete Florence1, Jonathan Tompson1, Jeannette Bohg2, and Debidatta Dwibedi1
Conference on Robot Learning (CoRL) 2021
1Robotics at Google,
2Stanford… See the full description on the dataset page: https://huggingface.co/datasets/Renton-Ren/multi-task-tcc-robosuite.robosuite-lift-pi05
Robosuite Lift — π0.5 zero-shot rollouts (1000 episodes)
1000 rollouts of the π0.5 policy (pi05_libero checkpoint) operating zero-shot on the
robosuite Lift task ("lift the red cube").
No fine-tuning was done on Lift — the LIBERO checkpoint is applied directly, since LIBERO is built
on robosuite and exposes an identical Panda + OSC_POSE interface.
Every rollout is kept, success or failure — each episode carries a success flag. Failures are
intentionally included (they are useful… See the full description on the dataset page: https://huggingface.co/datasets/jacob3333/robosuite-lift-pi05.plex_robosuite_rawVGCP_robosuite
VGCP: learning Inverse Dynamics Model using Inverse Reinforcement Learning
Preparation
1. Encoder for reward computation: The encoder is pretrained using TCC, and each is for one task. Download pretrain_runs.zip
2. Datasets for conditioning goals: The goals in observation are from a randomly sampled sequence in demonstrations. They are psuedo-goals. Download and put datasets under VGCP.
3. Finally, prepare video generative… See the full description on the dataset page: https://huggingface.co/datasets/Renton-Ren/VGCP_robosuite.eden_robosuiteSARM-robosuite-can-mh-stages
SARM Robosuite PickPlaceCan (Multi-Human) — Stage-Annotated Dataset
Stage-annotated demonstration dataset for the PickPlaceCan task from robosuite,
formatted in LeRobot v2.1 format for training
SARM (Stage-Aware Reward Models).
Dataset Summary
Property
Value
Task
Pick and place can
Robot
Panda (Franka Emika)
Environment
robosuite PickPlaceCan
Data source
robosuite multi-human (MH) demonstrations
Episodes
300
Total frames
62,756
FPS
20… See the full description on the dataset page: https://huggingface.co/datasets/poolvarine/SARM-robosuite-can-mh-stages.