aiqbench analysis
This research introduces AgilePE, a reinforcement learning system enabling autonomous UAVs to execute complex pursuit-evasion maneuvers through direct mapping of observations to control commands without intermediate planners. By utilizing competitive self-play and hardware-aligned simulations, the framework successfully transfers zero-shot policies from virtual environments to real quadrotors for rapid dodging and flanking tactics. The work advances robotics by demonstrating that sophisticated aerial strategies can emerge end-to-end via machine learning rather than traditional rule-based design.
AgilePE: Autonomous UAV Pursuit-Evasion via Self-Play Reinforcement Learning
arXiv:2608.14135v1 Announce Type: new Abstract: Autonomous pursuit-evasion is a fundamental challenge for Unmanned Aerial Vehicles (UAVs), requiring rapid decision-making under tightly coupled dynamics and continuously changing opponent behaviors. Traditional rule-based or differential-game approaches often struggle with high-dimensional aerial interactions and agile maneuvering. We present AgilePE, a complete system for autonomous UAV pursuit-evasion via self-play reinforcement learning. AgilePE integrates agile low-level control, competitive policy optimization, and sim-to-real deployment in a unified framework. The policy directly maps onboard state observations to Collective Thrust and Body Rates (CTBR) commands, enabling end-to-end agile maneuvering without intermediate trajectory planners or waypoint controllers. For training, we use competitive self-play with Prioritized Fictitious Self-Play (PFSP) and a diversified opponent pool, enabling agents to improve against historical policies while stabilizing optimization and reducing policy oscillation. This process leads to the emergence of sophisticated pursuit and evasion strategies. For real-world deployment, we develop a hardware-aligned simulation pipeline that models actuator-response dynamics, communication latency, and domain randomization. The learned policies transfer zero-shot to real quadrotors without task-specific tuning. Real-world experiments reproduce pursuit-evasion tactics observed in simulation, including rapid dodging and flanking, and demonstrate interactive two-agent zero-shot deployment.
Abstract:Autonomous pursuit-evasion is a fundamental challenge for Unmanned Aerial Vehicles (UAVs), requiring rapid decision-making under tightly coupled dynamics and continuously changing opponent behaviors. Traditional rule-based or differential-game approaches often struggle with high-dimensional aerial interactions and agile maneuvering. We present AgilePE, a complete system for autonomous UAV pursuit-evasion via self-play reinforcement learning. AgilePE integrates agile low-level control, competitive policy optimization, and sim-to-real deployment in a unified framework. The policy directly maps onboard state observations to Collective Thrust and Body Rates (CTBR) commands, enabling end-to-end agile maneuvering without intermediate trajectory planners or waypoint controllers. For training, we use competitive self-play with Prioritized Fictitious Self-Play (PFSP) and a diversified opponent pool, enabling agents to improve against historical policies while stabilizing optimization and reducing policy oscillation. This process leads to the emergence of sophisticated pursuit and evasion strategies. For real-world deployment, we develop a hardware-aligned simulation pipeline that models actuator-response dynamics, communication latency, and domain randomization. The learned policies transfer zero-shot to real quadrotors without task-specific tuning. Real-world experiments reproduce pursuit-evasion tactics observed in simulation, including rapid dodging and flanking, and demonstrate interactive two-agent zero-shot deployment.
8 pages, 7 figures. Under review
Robotics (cs.RO); Machine Learning (cs.LG)
arXiv:2608.14135 [cs.RO]
(or arXiv:2608.14135v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2608.14135 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 284 words
Related coverage
The Chase Is the Curriculum, the Capture Anchors the Credit: Pursuit-Evasion Self-Play for Zero-Data LLM Reasoning
arXiv · Aug 25, 2026
Deconfusing Corrigibility
AI Alignment Forum · Oct 17, 2026
What's Continual Learning, and Why Might We Expect To See It In Advanced LLM Agents?
LessWrong · Jun 12, 2026
Toward Training Superintelligent Software Agents through Self-Play SWE-RL
arXiv · Jun 3, 2026