This academic article focuses on modeling realistic joint constraints for human-robot interaction, biomechanics and robot-assisted rehabilitation. The authors propose a data-driven method to learn anatomically constrained upper-limb range of motion (RoM) boundaries from motion capture data. This is achieved using a one-class support vector machine with efficient hyper-parameter tuning. They also introduce an impairment index (II) metric for quantitative assessment of capability/impairment when comparing healthy and impaired arms. The study validates the metric on healthy subjects physically constrained to emulate hemiplegia and different disability levels as stroke patients.
Publication date: 20 Nov 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2311.10653
