aswinkumar99/LeRobot-SO101-Pick-Place

LeRobot数据集用于SO101拾取和放置任务，包含三种海绵操作任务和固定/随机物体布局。该数据集旨在用于模仿学习、行为克隆、策略评估以及布局变化下的鲁棒性研究。数据集结构分为三个任务组：单海绵无干扰物、多海绵无干扰物和单海绵有干扰物。每个任务组包含固定位置和随机位置两种布局。数据收集使用真实的SO101领导机器人，包括两个摄像头视角：头顶摄像头和手腕摄像头。任务提示为英文。数据集预期用途包括模仿学习、行为克隆、策略评估和位置变化的鲁棒性实验。

LeRobot dataset for SO101 pick-and-place with three sponge manipulation tasks and fixed/random object layouts. The dataset is intended for imitation learning, behavior cloning, policy evaluation, and robustness studies under layout changes. The dataset is organized into three task groups: single sponge with no distractors, multiple sponges with no distractors, and single sponge with distractors. Each task contains two location splits: fixed locations and random locations. The dataset was collected on a real SO101 leader robot, including two camera views: overhead camera and wrist camera. Task prompts are in English. The dataset is intended for imitation learning, behavior cloning, policy evaluation, and robustness experiments under location changes.