On Optimizing Interventions in Shared Autonomy

Authors: Weihao Tan, David Koleczek, Siddhant Pradhan, Nicholas Perello, Vivek Chettiar, Vishal Rohra, Aaslesha Rajaram, Soundararajan Srinivasan, H M Sajjad Hossain, Yash Chandak5341-5349

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

Reproducibility Variable Result LLM Response
Research Type Experimental To benchmark our algorithms, we conduct experiments using simulated human agents in the Lunar Lander (Brockman et al. 2016) and Super Mario Bros. (Kauten 2018) environments.
Researcher Affiliation Collaboration Weihao Tan1* , David Koleczek13 , Siddhant Pradhan1 , Nicholas Perello1, Vivek Chettiar2, Vishal Rohra2, Aaslesha Rajaram2, Soundararajan Srinivasan2, H M Sajjad Hossain2 , Yash Chandak 1 1 University of Massachusetts Amherst 2 Microsoft 3 Mass Mutual Data Science
Pseudocode Yes Algorithm 1: Hard Constrained Shared Autonomy
Open Source Code Yes Demo and code used for experiments can be found at https://github.com/David Koleczek/human_marl.
Open Datasets Yes Lunar Lander: A 2D simulation game from Open AI Gym (Brockman et al. 2016). Super Mario Bros. World 1-1: The first stage in the first world of the classic Super Mario game is implemented by the gym-super-mario-bros library (Kauten 2018).
Dataset Splits No The paper describes training agents and testing performance over 100 episodes but does not specify explicit train/validation/test dataset splits with percentages or counts, which is typical for static datasets rather than dynamic reinforcement learning environments.
Hardware Specification No The paper does not provide specific details about the hardware used for running its experiments, such as GPU/CPU models, memory, or cloud instance types.
Software Dependencies No The paper mentions using DDQN, SAC, Open AI Gym, and gym-super-mario-bros library but does not provide specific version numbers for these software components or any other ancillary software dependencies.
Experiment Setup No The paper states that 'A detailed listing of the hyperparameters used is located in the appendix(Tan et al. 2021)', meaning specific hyperparameter values are not provided directly in the main text.