Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Evaluating MEDIRL: A Replication and Ablation Study of Maximum Entropy Deep Inverse Reinforcement Learning for Human Social Navigation

Vinay Gupta, Nihal Gunukula

TL;DR

This work addresses reproducing MEDIRL for human social navigation in crowded environments and extends Fahad et al.’s original study through thorough ablations of learning parameters, state representations, network depth, and loss formulations. It demonstrates that a two-dimensional state representation can improve accuracy when vertical movement is absent, and highlights the critical role of the maximum-entropy component in maintaining exploratory, human-like navigation. The authors provide open resources via DagHub with data versioning and report practical lessons for robust HRI system development, including the importance of documentation and careful replication under hardware constraints. Overall, the study contributes to more reliable and socially aware navigation in human–robot interactions and guides future customization to different environmental contexts.

Abstract

In this study, we enhance the Maximum Entropy Deep Inverse Reinforcement Learning (MEDIRL) framework, targeting its application in human robot interaction (HRI) for modeling pedestrian behavior in crowded environments. Our work is grounded in the pioneering research by Fahad, Chen, and Guo, and aims to elevate MEDIRL's efficacy in real world HRI settings. We replicated the original MEDIRL model and conducted detailed ablation studies, focusing on key model components like learning rates, state dimensions, and network layers. Our findings reveal the effectiveness of a two dimensional state representation over three dimensional approach, significantly improving model accuracy for pedestrian behavior prediction in HRI scenarios. These results not only demonstrate MEDIRL's enhanced performance but also offer valuable insights for future HRI system development, emphasizing the importance of model customization to specific environmental contexts. Our research contributes to advancing the field of socially intelligent navigation systems, promoting more intuitive and safer human robot interactions.

Evaluating MEDIRL: A Replication and Ablation Study of Maximum Entropy Deep Inverse Reinforcement Learning for Human Social Navigation

TL;DR

This work addresses reproducing MEDIRL for human social navigation in crowded environments and extends Fahad et al.’s original study through thorough ablations of learning parameters, state representations, network depth, and loss formulations. It demonstrates that a two-dimensional state representation can improve accuracy when vertical movement is absent, and highlights the critical role of the maximum-entropy component in maintaining exploratory, human-like navigation. The authors provide open resources via DagHub with data versioning and report practical lessons for robust HRI system development, including the importance of documentation and careful replication under hardware constraints. Overall, the study contributes to more reliable and socially aware navigation in human–robot interactions and guides future customization to different environmental contexts.

Abstract

In this study, we enhance the Maximum Entropy Deep Inverse Reinforcement Learning (MEDIRL) framework, targeting its application in human robot interaction (HRI) for modeling pedestrian behavior in crowded environments. Our work is grounded in the pioneering research by Fahad, Chen, and Guo, and aims to elevate MEDIRL's efficacy in real world HRI settings. We replicated the original MEDIRL model and conducted detailed ablation studies, focusing on key model components like learning rates, state dimensions, and network layers. Our findings reveal the effectiveness of a two dimensional state representation over three dimensional approach, significantly improving model accuracy for pedestrian behavior prediction in HRI scenarios. These results not only demonstrate MEDIRL's enhanced performance but also offer valuable insights for future HRI system development, emphasizing the importance of model customization to specific environmental contexts. Our research contributes to advancing the field of socially intelligent navigation systems, promoting more intuitive and safer human robot interactions.
Paper Structure (19 sections, 6 equations, 1 figure, 1 algorithm)

This paper contains 19 sections, 6 equations, 1 figure, 1 algorithm.

Table of Contents

  1. Scope of Reproducibility
  2. Methodology
  3. Results
  4. Introduction
  5. Scope of Reproducibility
  6. Methodology
  7. Results
  8. Results reproducing original paper
  9. Discussion
  10. Appendix
  11. Additional Experimental Details
  12. Implementation
  13. Dataset
  14. Training Procedure
  15. Metrics
  16. ...and 4 more sections

Figures (1)

  • Figure 13: Figure displays the displacement of the predictions made of the model with mean squared instead of maximum entropy from the actual decisions made by the pedestrians.