Synthetic Parabolic Trajectories Dataset in Hexapod Robots

Parabolic trajectories constitute the primary motion of a robotic leg. Therefore, the data used to train models requires variations to the ideal (proposed) trajectories. Thus, it is essential to generate random sequences based on parabolic trajectories in each degree of freedom so that, collectively, they create a random three-dimensional step. The database comprises three classes of planar movements—two parabolic and one sigmoidal—representing one degree of freedom. Its structure consists of 200 timesteps, indicating the discretization of the continuous trajectory range into 1000 uniformly distributed trajectories across the three classes of movements in each degree of freedom.

抛物线轨迹是机器人腿部的主要运动模式。因此，用于模型训练的数据集需对理想（所提出的）轨迹进行变体生成。据此，需在每个自由度（degree of freedom）下基于抛物线轨迹生成随机序列，使各序列整体构成随机三维步态。该数据集涵盖三类平面运动——两类为抛物线轨迹运动，一类为Sigmoid轨迹运动——均对应单个自由度。该数据集的每条轨迹均包含200个时间步（timesteps），其构建方式为将连续轨迹区间离散为1000条均匀分布的轨迹，覆盖单个自由度下的三类平面运动。