Generating musculoskeletal gait trajectories that reflect patient‑specific parameters is a fundamental challenge for wearable robotics and rehabilitation. Assistive devices such as lower‑limb exoskeletons require reference trajectories that adapt to individual morphology and therapeutic goals while preserving biomechanical realism. Traditional solutions rely on handcrafted gait templates or computationally intensive optimization, which scale poorly across many subjects and walking conditions.
This work investigates conditional diffusion models for producing lower‑limb joint‑angle trajectories conditioned on gait parameters like step length. We implement two variants: a baseline transformer‑based diffusion model and a controllable diffusion transformer that incorporates adaptive normalization and classifier‑free guidance. Experiments on a dataset of 4,590 gait cycles demonstrate that diffusion models can synthesize realistic periodic gait trajectories and provide limited controllability over gait characteristics, highlighting their promise for personalized gait synthesis in assistive robotics.
Review