Motion Dataset for Homogeneous Characterization of Multi-Type Robots

This dataset employs a morphological data structure to achieve homogeneous robot characterization, establishing bidirectional labels between joint actuation and morphology. This dataset can be used for training a morphology prediction network based on joint-driven parameters (position *q*, velocity *qd*, torque *τ*). The dataset includes three types of robots: dual-arm, single-arm, and quadruped robots.