Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Offline Imitation of Badminton Player Behavior via Experiential Contexts and Brownian Motion

Kuang-Da Wang, Wei-Yao Wang, Ping-Chun Hsieh, Wen-Chih Peng

TL;DR

RallyNet addresses the challenge of imitating badminton players from offline data in a turn-based setting by modeling rally decision-making as a Contextual Markov Decision Process and introducing an Experiential Context Selector to encode player intent. It couples this with Latent Geometric Brownian Motion to capture interactions in latent space, along with an action projection layer and a joint loss that balances prediction, context learning, and stochastic dynamics. Trained on the largest real-world badminton dataset to date, RallyNet outperforms offline imitation baselines and state-of-the-art turn-based methods across multiple metrics, and offers actionable insights for coaching and sports analytics. This framework provides a principled, interpretable path to realistic rally generation and tactical interpretation in turn-based sports, with potential applicability beyond badminton.

Abstract

In the dynamic and rapid tactic involvements of turn-based sports, badminton stands out as an intrinsic paradigm that requires alter-dependent decision-making of players. While the advancement of learning from offline expert data in sequential decision-making has been witnessed in various domains, how to rally-wise imitate the behaviors of human players from offline badminton matches has remained underexplored. Replicating opponents' behavior benefits players by allowing them to undergo strategic development with direction before matches. However, directly applying existing methods suffers from the inherent hierarchy of the match and the compounding effect due to the turn-based nature of players alternatively taking actions. In this paper, we propose RallyNet, a novel hierarchical offline imitation learning model for badminton player behaviors: (i) RallyNet captures players' decision dependencies by modeling decision-making processes as a contextual Markov decision process. (ii) RallyNet leverages the experience to generate context as the agent's intent in the rally. (iii) To generate more realistic behavior, RallyNet leverages Geometric Brownian Motion (GBM) to model the interactions between players by introducing a valuable inductive bias for learning player behaviors. In this manner, RallyNet links player intents with interaction models with GBM, providing an understanding of interactions for sports analytics. We extensively validate RallyNet with the largest available real-world badminton dataset consisting of men's and women's singles, demonstrating its ability to imitate player behaviors. Results reveal RallyNet's superiority over offline imitation learning methods and state-of-the-art turn-based approaches, outperforming them by at least 16% in mean rule-based agent normalization score. Furthermore, we discuss various practical use cases to highlight RallyNet's applicability.

Offline Imitation of Badminton Player Behavior via Experiential Contexts and Brownian Motion

TL;DR

RallyNet addresses the challenge of imitating badminton players from offline data in a turn-based setting by modeling rally decision-making as a Contextual Markov Decision Process and introducing an Experiential Context Selector to encode player intent. It couples this with Latent Geometric Brownian Motion to capture interactions in latent space, along with an action projection layer and a joint loss that balances prediction, context learning, and stochastic dynamics. Trained on the largest real-world badminton dataset to date, RallyNet outperforms offline imitation baselines and state-of-the-art turn-based methods across multiple metrics, and offers actionable insights for coaching and sports analytics. This framework provides a principled, interpretable path to realistic rally generation and tactical interpretation in turn-based sports, with potential applicability beyond badminton.

Abstract

In the dynamic and rapid tactic involvements of turn-based sports, badminton stands out as an intrinsic paradigm that requires alter-dependent decision-making of players. While the advancement of learning from offline expert data in sequential decision-making has been witnessed in various domains, how to rally-wise imitate the behaviors of human players from offline badminton matches has remained underexplored. Replicating opponents' behavior benefits players by allowing them to undergo strategic development with direction before matches. However, directly applying existing methods suffers from the inherent hierarchy of the match and the compounding effect due to the turn-based nature of players alternatively taking actions. In this paper, we propose RallyNet, a novel hierarchical offline imitation learning model for badminton player behaviors: (i) RallyNet captures players' decision dependencies by modeling decision-making processes as a contextual Markov decision process. (ii) RallyNet leverages the experience to generate context as the agent's intent in the rally. (iii) To generate more realistic behavior, RallyNet leverages Geometric Brownian Motion (GBM) to model the interactions between players by introducing a valuable inductive bias for learning player behaviors. In this manner, RallyNet links player intents with interaction models with GBM, providing an understanding of interactions for sports analytics. We extensively validate RallyNet with the largest available real-world badminton dataset consisting of men's and women's singles, demonstrating its ability to imitate player behaviors. Results reveal RallyNet's superiority over offline imitation learning methods and state-of-the-art turn-based approaches, outperforming them by at least 16% in mean rule-based agent normalization score. Furthermore, we discuss various practical use cases to highlight RallyNet's applicability.
Paper Structure (20 sections, 17 equations, 5 figures, 2 tables)

This paper contains 20 sections, 17 equations, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Inverse Reinforcement Learning
  4. Offline Imitation Learning via Behavior Cloning
  5. Preliminaries
  6. Badminton: A Typical Example of a Turn-Based Sport
  7. The Contextual Markov Decision Process
  8. Problem Formulation
  9. Methodology
  10. Experiential Context Selector (ECS)
  11. Latent Geometric Brownian Motion (LGBM)
  12. Action Projection Layer
  13. Loss Function
  14. Experiments
  15. Experimental Setup
  16. ...and 5 more sections

Figures (5)

  • Figure 1: An illustration of RallyNet in a badminton rally.
  • Figure 2: The framework of RallyNet.
  • Figure 3: The length distribution difference. The orange distribution represents the ground truth rally length distribution, while the blue distribution depicts the generated rally length distribution.
  • Figure 4: The landing distributions of the player in various states are shown, with RallyNet in the upper row and BC in the lower row.
  • Figure 5: Interpreting the intent behind an action. player A's clear stroke is meant to plan for a smash.