Semi-physical Simulation System Of Operation Strategy For Supernumerary Robotic Limbs

Unlike exoskeleton robots that enhance human limbs,supernumerary robotic limbs are worn as an extension of human limbs to assist wearers in completing specific tasks.This new type of wearable robot has been proposed in recent years.In task scenarios where the work is of high intensity and subject to space and scale constraints,such as aircraft manufacturing and building decoration,supernumerary robotic limbs can be docked and integrated with human limbs to mutually assist and cooperate,thereby improving the wearer’s perception and operational capabilities and completing complex tasks.As supernumerary robotic limbs require close collaboration with their wearers during operation and often face hazardous working conditions,ensuring wearer safety is of utmost importance.To this end,this thesis proposes a semi-physical simulation system for supernumerary robotic limbs operation strategy,which can be used to verify algorithms in the early stages of a project and conduct motion simulation analysis under extreme conditions.This provides a new approach to improving the efficiency of wearable robot-assisted work.The main research contents of this thesis are as follows:(1)According to the system requirements,a dynamic model of the supernumerary robotic limbs and the wearer that meets the needs of the task is constructed as the basis for verifying the subsequent control algorithm.Firstly,the structure of the supernumerary robotic limbs used is introduced,and according to the connection characteristics between the components,appropriate dynamic constraints are established between the components through the RecurDyn software,and the dynamics simulation model of the wearer-external limb robot is constructed.Finally,based on the teaching action,verifying the accuracy of the established dynamic model.(2)Aiming at the "gradient explosion" problem in the training process of deep network,a deep residual GRU neural network model is proposed for wearer behavior recognition.Firstly,the motion information of the wearer is captured through the inertial measurement unit,and experiments are designed to collect and produce a data set of the wearer’s behavior;then,the data is processed by filtering,segmentation,normalization and other operations to remove the noise components in the data and speed up the convergence of later model training speed and improve the recognition accuracy;finally,by introducing the commonly used recursive loop network unit and residual learning strategy,a deep residual GRU network is constructed to classify the wearer’s behavior,and the recognition accuracy rate is 92.20%,which is verified by comparative experiments,superior performance of the model.(3)According to the multi-working conditions and multi-constraint characteristics of the supernumerary robotic limbs,the single-arm motion of the supernumerary robotic limbs is planned based on the quintic polynomial interpolation algorithm.First,according to the characteristics of the single-arm connection of the supernumerary robotic limbs,establish the corresponding link coordinate system,and use the D-H parameter method to derive its kinematic equation;through Matlab programming,compare the interpolation algorithm characteristics and actual interpolation effects of cubic polynomial and quintic polynomial,and finally adopt the quintic polynomial interpolation algorithm performs trajectory planning on the end of the external limb robot to complete the collaborative task with the wearer.(4)Aiming at the task requirements of the supernumerary robotic limbs,a semi-physical simulation system for the operation strategy of the supernumerary robotic limbs is constructed to complete a variety of task simulations.Through the analysis of the system framework,the role of each module in the system operation process is introduced,and the simulation analysis of auxiliary handling,auxiliary support,support unloading and other operation tasks is completed by using RecurDyn software,and the auxiliary operation mechanism and tasks characteristics in different operation scenarios are revealed,which have guiding significance for practical engineering.