Learned Controllers for Agile Quadrotors in Pursuit-Evasion Games

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Roncero, Alejandro Sanchez, Cai, Yixi, Andersson, Olov, Ogren, Petter
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915494606405632
author Roncero, Alejandro Sanchez
Cai, Yixi
Andersson, Olov
Ogren, Petter
author_facet Roncero, Alejandro Sanchez
Cai, Yixi
Andersson, Olov
Ogren, Petter
contents We address the problem of agile 1v1 quadrotor pursuit-evasion, where a pursuer and an evader learn to outmaneuver each other through reinforcement learning (RL). Such settings face two major challenges: non-stationarity, since each agent's evolving policy alters the environment dynamics and destabilizes training, and catastrophic forgetting, where a policy overfits to the current adversary and loses effectiveness against previously encountered strategies. To tackle these issues, we propose an Asynchronous Multi-Stage Population-Based (AMSPB) algorithm. At each stage, the pursuer and evader are trained asynchronously against a frozen pool of opponents sampled from a growing population of past and current policies, stabilizing training and ensuring exposure to diverse behaviors. Within this framework, we train neural network controllers that output either velocity commands or body rates with collective thrust. Experiments in a high-fidelity simulator show that: (i) AMSPB-trained RL policies outperform RL and geometric baselines; (ii) body-rate-and-thrust controllers achieve more agile flight than velocity-based controllers, leading to better pursuit-evasion performance; (iii) AMSPB yields stable, monotonic gains across stages; and (iv) trained policies in one arena size generalize fairly well to other sizes without retraining.
format Preprint
id arxiv_https___arxiv_org_abs_2506_02849
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learned Controllers for Agile Quadrotors in Pursuit-Evasion Games
Roncero, Alejandro Sanchez
Cai, Yixi
Andersson, Olov
Ogren, Petter
Robotics
Machine Learning
We address the problem of agile 1v1 quadrotor pursuit-evasion, where a pursuer and an evader learn to outmaneuver each other through reinforcement learning (RL). Such settings face two major challenges: non-stationarity, since each agent's evolving policy alters the environment dynamics and destabilizes training, and catastrophic forgetting, where a policy overfits to the current adversary and loses effectiveness against previously encountered strategies. To tackle these issues, we propose an Asynchronous Multi-Stage Population-Based (AMSPB) algorithm. At each stage, the pursuer and evader are trained asynchronously against a frozen pool of opponents sampled from a growing population of past and current policies, stabilizing training and ensuring exposure to diverse behaviors. Within this framework, we train neural network controllers that output either velocity commands or body rates with collective thrust. Experiments in a high-fidelity simulator show that: (i) AMSPB-trained RL policies outperform RL and geometric baselines; (ii) body-rate-and-thrust controllers achieve more agile flight than velocity-based controllers, leading to better pursuit-evasion performance; (iii) AMSPB yields stable, monotonic gains across stages; and (iv) trained policies in one arena size generalize fairly well to other sizes without retraining.
title Learned Controllers for Agile Quadrotors in Pursuit-Evasion Games
topic Robotics
Machine Learning
url https://arxiv.org/abs/2506.02849