A Dataset for Gait Analysis at Different Speeds Using Motion Capture, Instrumented Treadmill and IMUs

We present a comprehensive gait dataset from 16 able-bodied participants walking under diverse conditions, to enhance data-driven models for real-world gait analysis. Utilizing marker-based motion capture, inertial measurement units, and an instrumented treadmill, we captured lower limb kinematics during treadmill and overground walking, and kinetics during treadmill walking. The dataset encompasses speeds from 0.1 to 2 m\/s, variable step lengths, high-acceleration transitions, stop-and-go sequences,  turning maneuvers reflecting self-directed walking patterns, and asymmetric gait patterns like limping induced by a split-belt treadmill, simulating certain pathological gait such as in hemiparesis.  The comprehensive nature of this dataset enables the development of robust models that reflect the variability and complexity inherent in natural walking. Bridging the gap between controlled lab settings and real-world conditions, the dataset enables the creation of predictive models that are accurate and generalizable. This resource advances tools for diagnostics, monitoring, rehabilitation, and assistive technologies such as lower-body exoskeletons, ultimately contributing to better clinical outcomes  and a deeper understanding of human gait.