This article focuses on the development and evaluation of a patient behaviour model, which is to be used to quantify the patient’s contribution to the generation of the movements performed during robot-assisted lower limb mobilization. This model allows adaptation of the robot’s behaviour based on the patient’s capabilities in order to guarantee a therapy progression. The work provides a strategy based both on first principles and data-driven approaches, to ensure the trustworthiness of the model (due to knowledge of the underlying mechanics) and suitability of the model for limited amounts of data. Our best performing model, which includes a cooperation between those two strategies, is able to predict the required torques for the mobilization of a completely passive subject during different therapy configurations. The model only uses data available from the robotic system without exploiting additional sensors. Validation is conducted with real human data from a total of 10 healthy subjects.
Soavi G., Knopp T., Borner H., Hirt F., Hildenbrand X., Konig A., et al. (2024). Development of a Patient Behaviour Model for an Adaptive Medical Robot. GEWERBESTRASSE 11, CHAM, CH-6330, SWITZERLAND : Springer Nature [10.1007/978-3-031-55000-3_4].
Development of a Patient Behaviour Model for an Adaptive Medical Robot
Diversi R.;
2024
Abstract
This article focuses on the development and evaluation of a patient behaviour model, which is to be used to quantify the patient’s contribution to the generation of the movements performed during robot-assisted lower limb mobilization. This model allows adaptation of the robot’s behaviour based on the patient’s capabilities in order to guarantee a therapy progression. The work provides a strategy based both on first principles and data-driven approaches, to ensure the trustworthiness of the model (due to knowledge of the underlying mechanics) and suitability of the model for limited amounts of data. Our best performing model, which includes a cooperation between those two strategies, is able to predict the required torques for the mobilization of a completely passive subject during different therapy configurations. The model only uses data available from the robotic system without exploiting additional sensors. Validation is conducted with real human data from a total of 10 healthy subjects.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
