Design And Analysis Of A Cable-Driven Parallel Mechanism For Jaw Movement

Mandibular robots are widely used in dental science,food science,biomechanics and medical rehabilitation.Aiming at the shortage of bionics of existing mandibular robots,in order to improve the bionics of mandibular-like mechanism and to reproduce human mandibular motion more truly,a mandibular-like motion mechanism based on cable redundancy is proposed in this paper,which is based on the redundant driving characteristics of human stomatognathic system,the variable stiffness characteristics of mandibular motion,the unidirectional pulling characteristics of human muscles and the attachment position of masticatory.Firstly,according to the bionic mechanism of human mandible,three groups of masticatory muscles(masseter,temporalis and lateral pterygoid muscles)are simulated by cable traction,and the human temporomandibular joint(TMJ)is simulated by point-contact high kinematic pair(HKP).Based on the anatomical parameters of human mandible,the connecting position between cable and upper and lower platforms and the high pair structure of point-contact are designed.Pneumatic artificial muscle is selected as the actuator of the mechanism,and the overall design of the system is completed.The mechanism has the characteristics of redundant drive,variable stiffness and compact structure.Secondly,the degree of freedom analysis,inverse kinematics solution and Jacobian matrix derivation of the cable-driven parallel mechanism for jaw movement are carried out.To study the relationship between the unidirectional force characteristics of the cable and its configuration,the closed space solution and analysis of the mechanism are carried out.The virtual prototype model of the mechanism was established by using Adams,a multi-body system dynamics simulation platform,and the trajectory planning and motion simulation of mandibular functional motion(open-close motion,forward-backward motion,lateral motion)were carried out.The simulation results show that the change of cable length is similar to the contraction of human masticatory muscles,which plays an important role in studying the contraction and stress of masticatory muscles during functional mandibular movement.Then,the first kind of Lagrange equation is used to model the dynamics of the mandibular-like mechanism.The optimal allocation of driving force is solved by the two optimization objectives of the minimum L1 norm and L2 norm of the cable driving power respectively.A method of optimal allocation of driving force based on projection neuralnetwork is proposed,which transforms the optimal allocation of driving force into the projection neurodynamics problem.The numerical results show that the maximum instantaneous driving pull and the maximum instantaneous driving power can be reduced by44% and 36% respectively with the minimum L2 norm of the rope driving power as the optimization objective,which can improve the mechanical properties and load-carrying capacity of the mechanism.To solve the problem of force optimal allocation in this paper,the projection neural network method is better than the conventional optimization method(interior point method,sequential quadratic programming and active set method)has higher computational efficiency.and with the development of parallel computing,it has more extensive application prospects.Finally,three control laws,PD control,augmented PD control and adaptive control based on RBF network approximation,are designed for the mandibular mechanism.The design process of RBF network and adaptive control rate based on RBF network approximation is briefly described.Under the dynamic equation of mandibular mechanism described by S function,two control laws,PD control and augmented PD control,are used respectively;under the virtual prototype model of mechanism established in Adams,augmented PD control and adaptive control based on RBF network approximation are used respectively.Two sets of simulation results show that under the precise dynamic modeling,the tracking error of the augmented PD control is smaller than that of the PD control,and the driving pull of the cable is better,and that under the imprecise dynamic modeling,the tracking error of the adaptive control based on RBF network approximation is better than that of the augmented PD control generally,but in the initial stage of simulation,the cable-driven pull jitter is more obvious.