D-Reflex Dataset

该数据集包含了模拟人形机器人与墙壁互动以避免跌倒的场景，涵盖了墙壁距离、机器人姿态以及损伤组合的各种配置。此外，该数据集还包括接触映射，并允许对机器人与墙壁互动的左右两侧进行分析。该数据集的规模达到了882,000次模拟，其任务是学习如何在人形机器人上选择接触墙面的位置，以防止跌倒。

This dataset encompasses scenarios where humanoid robots interact with walls to avoid falls, covering various configurations of wall distances, robot postures, and damage combinations. Furthermore, the dataset includes contact mapping and allows analysis of both left and right sides of the robot's interaction with walls. The dataset consists of 882,000 simulation trials, with the objective of learning how to select appropriate wall-contact positions on humanoid robots to prevent falls.