Lower-limb joint and Cartesian gait trajectory dataset with rehabilitation-oriented step length and foot clearance ranges: real and synthetically generated data

Key contribution: to the authors' knowledge, no public dataset currently exists with real motion-capture gait trials covering step length and foot clearance ranges tailored to rehabilitation needs (i.e. extending beyond typical healthy-gait parameters). This dataset fills that gap and additionally provides the full pipeline used to generate and validate synthetic gait trajectories from it. Contains the complete pipeline for generating and validating lower-limb joint and Cartesian gait trajectories for a lower-limb exoskeleton: (1) VICON motion capture of 14 subjects used to design the kinematic model, (2) VICON motion capture of 8 subjects reserved for validation, across multiple speeds, step lengths and foot clearance heights, (3) validation of an Invasive Weed Optimization (IWO) inverse kinematics algorithm against the 8 validation subjects, (4) 1487 synthetic gait trajectory samples generated by the validated algorithm and used to train a neural network, and (5) final validation of the trained neural network (as real-time implementation of the algorithmic generator) against the 8 validation subjects. See README.md for full documentation of file structure, naming conventions, units, and subject identification.