PlanGAN: Model-based Planning With Sparse Rewards and Multiple Goals
Authors: Henry Charlesworth, Giovanni Montana
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The performance of Plan GAN has been tested on a number of robotic navigation/manipulation tasks in comparison with a range of model-free reinforcement learning baselines, including Hindsight Experience Replay. We perform experiments in four continuous environments built in Mu Jo Co [39] Four Rooms Navigation, Three-link Reacher, Fetch Push and Fetch Pick And Place (see Figure 3). Full details about these environments, along with the hyper parameters used for the experiments, can be found in the Appendix. |
| Researcher Affiliation | Academia | Henry Charlesworth Giovanni Montana Warwick Manufacturing Group University of Warwick Coventry, United Kingdom {H.Charlesworth, G.Montana}@warwick.ac.uk |
| Pseudocode | Yes | Algorithm 1: Plan GAN |
| Open Source Code | No | The paper provides a link for 'Videos of results' (https://sites.google.com/view/plangan/home) which is a project homepage, but it does not contain an unambiguous statement that the source code for the methodology is openly available or provide a direct link to a code repository. |
| Open Datasets | Yes | We perform experiments in four continuous environments built in Mu Jo Co [39] Four Rooms Navigation, Three-link Reacher, Fetch Push and Fetch Pick And Place (see Figure 3). |
| Dataset Splits | No | The paper describes evaluating performance over time steps of interaction but does not specify explicit train/validation/test dataset splits as would be typical for static datasets. It mentions 'hyper parameters used for the experiments, can be found in the Appendix.' but not how data for training, validation, or testing is partitioned. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU or CPU models, memory, or cloud instance types used for running the experiments. |
| Software Dependencies | No | The paper mentions software like MuJoCo [39] and Open AI Baselines [40] but does not provide specific version numbers for these or any other ancillary software components, which are required for reproducible descriptions. |
| Experiment Setup | Yes | Full details about these environments, along with the hyper parameters used for the experiments, can be found in the Appendix. |