Research On The Control Of Artificial Muscles Based On Stochastic Recruitment

With the development of robotics,artificial muscle has gradually become a research hotspot in actuator field.Compared with traditional actuators,artificial muscles have advantages such as high energy-dense,vast degree of freedom,no noise,etc.However,artificial muscles are hard to control since they have pronounced hysteresis.Stochastic recruitment is a novel control method for artificial muscles.This control architecture is inspired by skeletal muscles,it divides artificial muscle into many tiny units and control the collective behavior of these units.Researches in this thesis focus on stochastic recruitment control.Major findings include four hands as follows:Firstly,hysteresis loop exists in artificial muscles is studied,and is modeled as binary state Markov cell.System consists of binary state cells is studied,and close-loop,open-loop and optimal stochastic recruitment control law are analyzed.Stochastic recruitment simulation platform is founded in Simulink environment,and three control laws are validated;Secondly,binary state Markov cell is generalized to multi-state situation,for multi-state cell artificial muscle,stability of stochastic recruitment control is analyzed,and corresponding close-loop,open-loop and optimal stochastic recruitment control law is proposed.These three control laws are validated in Simulink environment;Thirdly,nonuniform cells in stochastic recruitment is studied,and stochastic recruitment control law is extended to nonuniform situation.Nonuniform in transition probabilities is analyzed,and stochastic recruitment control law with nonuniform in cell length is derivated.Stochastic recruitment control law with two kinds of nonuniformity is proposed and validated.Besides,nonuniform stochastic recruitment control law is designed to minimize the number of cells to generate desired output;Finally,three dynamic performance indexes of control system are chosen as measurement.Partial stochastic recruitment control and stochastic recruitment fuzzy control are proposed,and relationship between cell failure and overshoot in stochastic recruitment is analyzed.Dynamic performance in stochastic recruitment control is improved in above three different ways.