Comp-ACT Evaluation Dataset

该数据集包含了在双臂和单臂机器人设置下，针对各种复杂接触丰富操作任务进行的Comp-ACT和ACT方法的演示和性能评估。数据来源于两台配备内置力矩传感器的UR5e机器人手臂，以及用于任务观察的各种相机设置。该数据集在6项任务中进行评估（1项模拟任务，5项现实世界任务），旨在对接触丰富操作任务的评估进行研究。This dataset contains demonstrations and performance evaluations of the Comp-ACT and ACT methods for a wide range of complex contact-rich manipulation tasks under both dual-arm and single-arm robotic setups. The data is collected from two UR5e robotic arms equipped with built-in torque sensors, along with various camera configurations for task observation. This dataset is evaluated across six tasks, including one simulated task and five real-world tasks, and is designed to support research on the evaluation of contact-rich manipulation tasks.