On-demand auxiliary rehabilitation robot training method based on Bayesian optimization

Abstract

An on-demand auxiliary rehabilitation robot training method based on Bayesian optimization comprises the steps that performance indexes of a user in a random preset track following task for multiple times are collected, performance of the user in each round is evaluated in a cost function mode, and in each subsequent round of task, hyper-parameter optimization of the on-demand auxiliary algorithm is performed based on the Bayesian optimization algorithm and the previous task performance of the user, and the on-demand auxiliary effect of the next task is adjusted. Due to the fact that exercise ability differences of different subjects are large, learning ability in the training process are different, the exercise ability of a user cannot be self-adapted through a traditional training mode, and in order to improve the rehabilitation training effect of the subjects and improve the adaptability of rehabilitation strategies to different subjects. According to the method, the variability of the task can be effectively improved, the adaptability of the equipment to the exercise ability of the user can be more quickly realized through a shorter cycle process, the workload of a therapist is reduced, and the rehabilitation efficiency is effectively improved.

Description

technical field [0001] The present invention relates to rehabilitation robots, rehabilitation training and machine learning technical fields, in particular to an on-demand auxiliary rehabilitation robot training method based on Bayesian optimization. Background technique [0002] According to statistics, in the past 30 years, the incidence of stroke in my country has continued to increase. With the aging of society and the acceleration of urbanization, the incidence of stroke has risen sharply. At present, stroke accounts for more than 20% of the death rate of residents in my country. More than 75% of survivors suffer from motor dysfunction such as hemiplegia, and limb dysfunction will directly affect the daily life activities of patients. A large number of clinical tests have shown that exercise rehabilitation therapy, that is, the rehabilitation physicians assist the affected limb to move for many times for a long time, has a positive effect on the rehabilitation of the pat...

AI Technical Summary

Problems solved by technology

The present invention monitors the sports performance of different subjects, can realize personalized adaptive adjustment of training tasks for each subject, and can reduce the boring and cumbersome work of therapists / technicians adjusting task difficulty for different subjects in real time In the patent "Bayesian Optimization-Based Active Training Reference Trajectory Complexity Adjustment Method for Upper Limb Rehabilitation Robot", the Bayesian optimization method is used to adjust the complexity of the training reference trajectory, without consideration The case where the designed reference trajectory is fixed

And when the designed reference trajectory is fixed, what is adjusted is only the stance control parameters of on-demand assistance, and the results of any experimental process will not have any impact on the trajectory, and it does not mention the content related to on-demand assistance. The form and calculation method of the auxiliary force field control is an important point. The application stated that the training of the subjects was based on the form of teaching and active training, and the trajectory complexity was used to optimize the target. It did not consider the basis of Based on the trajectory drawing of the subjects in the auxiliary force field, the user's trajectory following error and the work done during the process are used as evaluation indicators, and it is not considered to optimize the strength of the auxiliary force field based on these evaluation indicators

Images