RESUMO
We developed a flexible support with embedded polymer optical fiber (POF) sensors for the assessment of humanâ»robot interaction forces. The supports were fabricated with a three-dimensional (3D) printer, where an acrylonitrile butadiene styrene (ABS) rigid structure was used in the region of the support in which the exoskeleton was attached, whereas a thermoplastic polyurethane (TPU) flexible structure was printed in the region where the users placed their legs. In addition, fiber Bragg gratings (FBGs), inscribed in low-loss, cyclic, transparent, optical polymer (CYTOP) using the direct-write, plane-by-plane femtosecond laser inscription method, were embedded in the TPU structure. In this case, a 2-FBG array was embedded in two supports for humanâ»robot interaction force assessment at two points on the users' legs. Both FBG sensors were characterized with respect to temperature and force; additionally, the creep response of the polymer, where temperature influences the force sensitivity, was analyzed. Following the characterization, a compensation method for the creep and temperature influence was derived, showing relative errors below 4.5%. Such errors were lower than the ones obtained with similar sensors in previously published works. The instrumented support was attached to an exoskeleton for knee rehabilitation exercises, where the humanâ»robot interaction forces were measured in flexion and extension cycles.
RESUMO
This paper demonstrates the use of polymer optical fiber Bragg gratings (POFBGs) for angle measurements over a range of different oscillatory frequencies. The POFBGs are inscribed in low-loss, cyclic transparent amorphous fluoropolymers (CYTOP) and are imprinted using the direct-write, plane-by-plane femtosecond laser inscription method. As the polymer has a viscoelastic response and given that the Young's modulus depends on the oscillatory frequency, a compensation technique for sensor frequency cross-sensitivity and hysteresis is proposed and verified. Results show that the proposed compensation technique is able to provide a root mean squared error (RMSE) reduction of 44%, and a RMSE as low as 2.20° was obtained when compared with a reference potentiometer. The hysteresis reduction provided by the proposed technique is 55%, with hysteresis <0.01. The results presented in this paper can pave the way for movement analysis with POFBG providing higher sensitivity and low hysteresis over a large range of motion frequencies.