TarGF: Learning Target Gradient Field to Rearrange Objects without Explicit Goal Specification

Authors: Mingdong Wu, Fangwei Zhong, Yulong Xia, Hao Dong

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results in ball rearrangement and room rearrangement demonstrate that our method significantly outperforms the state-of-the-art methods in the quality of the terminal state, the efficiency of the control process, and scalability.
Researcher Affiliation Academia Mingdong Wu* 1, 3, Fangwei Zhong* 2, 3, Yulong Xia1, Hao Dong1, 4 1 Center on Frontiers of Computing Studies, School of Computer Science, Peking University 2 School of Intelligence Science and Technology, Peking University 3 Beijing Institute for General Artificial Intelligence (BIGAI) 4 Peng Cheng Laboratory {wmingd, zfw, hao.dong}@pku.edu.cn
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes The code and demo videos are on https://sites.google.com/view/targf.
Open Datasets Yes Room Rearrangement is built on a more realistic dataset, 3D-FRONT [46].
Dataset Splits No The paper specifies training and testing splits for the Room Rearrangement dataset (756 for training, 83 for testing) but does not explicitly mention a 'validation' split.
Hardware Specification No The main paper states that hardware details are provided in the supplemental material, but does not include them in the paper itself.
Software Dependencies No The paper cites external tools like 'Pybullet' and 'Soft actor-critic (sac) implementation in pytorch' but does not list specific version numbers for its own software dependencies.
Experiment Setup No The main paper states that training details and hyperparameters are provided in the supplemental material, but does not include them in the paper itself.