arXiv Game AI· Jiqing Wen·· 4 d agoAI score38
Learned Combat Skill Motion Generation Framework for VR Fighting Games
Learning Martial Arts Motion Skills for Animating Opponents in VR Sparring Games
AI summary
The research proposes a combat skill learning framework that pretrains a reusable combat skill library from taekwondo motion capture data via reinforcement learning, and adapts it through adversarial style rewards and task-specific rewards to generate physically plausible and technically well-formed virtual opponent motions. A user study verified the physical fidelity, smoothness, and martial arts form quality of the generated motions, and a prototype system supports interactive opponent behavior on Meta Quest headsets, demonstrating the feasibility of the learned skill library for VR fighting game animation.
Source: arXiv Game AI · arxiv.org