WILD-SCAV

WILD-SCAV是一个基于Unity3D的3D开放世界FPS游戏环境，由Inspir.AI开发。该数据集提供了一个高度可配置的环境，包括多种任务和交互模式，使AI代理能够学习感知3D环境、导航和规划、竞争和合作等能力。WILD-SCAV支持不同复杂度，如可配置的地图、建筑结构和分布，以及多代理设置，包括合作和竞争任务。该数据集主要用于评估和推动强化学习算法的发展，特别是在开放世界环境中的应用，旨在解决复杂任务解决能力的挑战。

WILD-SCAV is a 3D open-world FPS game environment based on Unity3D, developed by Inspir.AI. This dataset provides a highly configurable environment encompassing diverse tasks and interaction modes, enabling AI Agents to learn capabilities such as 3D environment perception, navigation and planning, as well as competition and collaboration. WILD-SCAV supports varying complexity levels, including configurable maps, architectural structures and their distributions, as well as multi-agent settings covering both collaborative and competitive tasks. This dataset is primarily used to evaluate and advance the development of reinforcement learning algorithms, especially for applications in open-world environments, aiming to address the challenges of complex task-solving capabilities.