M3Bench
收藏资源简介:
M3Bench是由华中科技大学和北京通用人工智能研究所联合开发的一个全面的全身运动生成基准数据集,旨在促进家庭环境中移动操作的研究。该数据集包含30,000个物体重新排列任务,跨越119个不同的家庭场景,涵盖了广泛的环境和任务目标约束。数据集通过M3BenchMaker工具自动生成,该工具从高级任务指令生成协调的全身运动轨迹,仅需要基本的场景和机器人信息。M3Bench的应用领域包括机器人学中的全身运动生成,旨在解决在复杂3D环境中移动操作的协调问题。
M3Bench is a comprehensive full-body motion generation benchmark dataset jointly developed by Huazhong University of Science and Technology and Beijing Institute for General Artificial Intelligence, with the goal of advancing research on mobile manipulation in home environments. It encompasses 30,000 object rearrangement tasks across 119 unique home scenarios, covering a broad spectrum of environmental and task goal constraints. The dataset is automatically generated via the M3BenchMaker tool, which generates coordinated full-body motion trajectories from high-level task instructions, requiring only basic scene and robot configuration information. The application domains of M3Bench include full-body motion generation in robotics, aiming to address the coordination challenges of mobile manipulation in complex 3D environments.

- 1M${}^{3}$Bench: Benchmarking Whole-body Motion Generation for Mobile Manipulation in 3D Scenes华中科技大学电子信息与通信学院 · 2024年



