Research On Human-computer Interaction System Based On Lower-limb Musculoskeletal Model

For the past few years,with the population of the aged,the arthritis,the strokes and other patients continuing to grow,the intelligent rehabilitation robots have received more and more attention and has gradually become an important tool which can replace the traditional physiotherapists to assist rehabilitation training.In the field where the robots assist people to recover through exercise,in order to achieve patient-dominated,natural and intelligent robot control,the control of the robots should take the patient's active movement intentions and the body's activity state into account.Through the human body's musculoskeletal model to create human-computer interaction interface and system,it can provide a kind of autonomous and flexible control means for the rehabilitation robots,in the premise of ensuring the subjectivity and the flexibility of the human body to improve the patients' rehabilitation effect.In this paper,the object is the movement part of human joint——the skeletal muscles.Combining with the analysis and kinematics information feedback of the surface EMG signal in movement process,we study the biomechanics process generated by the human body's motion,and establish a new musculoskeletal model of human's lower limb.Then based on the model we design and implement an intuitive and natural,flexible and efficient,safe and reliable human-computer interaction system.The main research work of this paper includes:(1)In order to improve the accuracy of the musculoskeletal model,aiming at the individual differences of the muscle activation,researching the method of extracting the human body's surface EMG signal' characteristic which represents the body's activities,the mapping relation between EMG signal and force information and the correlation between EMG parameters and joint angle were analyzed by experiment.On this basis,select the appropriate EMG eigenvalues as the input of the musculoskeletal model,through an experiment to verify and analyze the characteristic parameters in the actual environment.(2)Aiming at the problem that the lack of the autonomy and the accuracy of the human-computer interaction interface,an improved modeling method of the lower limb's skeletal muscle is proposed to obtain more accurate control information.Introducing the Hill model to estimate the knee related muscle force when exercising,a parallel cascade identification(PCI)model was established to researching the mapping relationship between EMG signal and muscle force,and then the fusion of the two to establish a new Hill-PCI force prediction model to achieve an accurate estimate of muscle force.In order to obtain more polybasic control information,researching the proportion model about myoelectricity angle prediction,through the nonlinear iterative partial least squares method to establish the correspondence between the EMG eigenvalue and the joint angle to improve the accuracy of the angle prediction.At last,the feasibility and the effectiveness of the whole model are verified by experiments.(3)Designing and developing a human-computer interaction system based on the musculoskeletal model,according to the muscle force and joint angle information which the model predicts,formulating reasonable interactive control instructions to realize the direct control of the human body's lower limbs to virtual objects.In the actual subjects,combining the lower limbs' musculoskeletal model and the human body's motion parameters,to realize a rehabilitation training system based on the interactive control of the subjects and the virtual objects,to promote the effect of muscle's strength training and to enhance the intersetingness of the training.