Box-Packing Task Simulation
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/AssassinWS/LLM-TAMP
下载链接
链接失效反馈官方服务：
资源简介：
该数据集基于PyBullet模拟环境，针对桌面装箱任务进行设计，旨在评估LLM3框架的有效性。数据集包含了不同配置以及来自消融研究的比较结果，这些研究对比了不同的LLM3策略及其在规划成功率以及运动规划器调用次数方面的有效性。该数据集涵盖了两种设置，分别是桌面装箱任务，其中涉及不同物体尺寸和碰撞情景的变体。所涉及的任务为任务与动作规划（TAMP）。
This dataset is developed based on the PyBullet simulation environment, tailored for the desktop bin packing task, with the objective of evaluating the effectiveness of the LLM3 framework. It includes diverse configurations and comparative results from ablation studies, which compare different LLM3 strategies and their effectiveness in terms of planning success rate and the number of motion planner invocations. The dataset covers two settings, both within the desktop bin packing task, involving variants with different object sizes and collision scenarios. The addressed tasks belong to Task and Motion Planning (TAMP).
提供机构：
Authors of the paper
