Research On Limit Cycle Walking Of Passive Biped Robot

Compared with general robots,biped robots have better environmental adaptability and flexibility,and are relatively easier to walk on some complex terrains,such as mountain jungles,etc.,so they have received extensive attention in military,entertainment and other fields.The passive walking research initiated by Mc Geer provides a new direction for the biped robot walking research.The passive walking machine can walk on a slope without relying on external energy input and relying on gravity alone.Compared with traditional robot walking,it has More anthropomorphic gait and higher energy efficiency.This subject mainly hopes to be able to study the mechanism of passive walking,and provide relevant theoretical groundwork for adding active control on the basis of passive walking in the future.This research refers to the research of classic passive biped robots at home and abroad,and simplifies some conditions to establish models with straight legs and knees.The walking process is established as two stages of continuous swing and discrete collision.For the swing phase,the Lagrangian method is used to analyze the dynamics of the two structures during walking;for the discrete collision phase,the angular momentum conservation relationship is used to analyze the velocity before and after the collision.The corresponding relationship lays a theoretical foundation for the study of the limit cycle characteristics of passive walking and the study of active and active drive control.Limit cycle walking means that it is not stable at every moment in the walking process,but the overall walking has periodicity and stability.However,limit cycle walking is very sensitive to initial conditions,and continuous walking must be achieved under appropriate initial conditions.Therefore,on the basis of the above dynamic model,this paper uses search algorithms such as Newton's method and genetic algorithm to find the fixed points and limit cycles of passive walking under given conditions,which shows the effectiveness of these search algorithms for limit cycle search,and The stability and sensitivity of the fixed points found are analyzed.On the basis of the above theoretical analysis model,this paper also uses virtual prototype simulation and actual physical prototype experiment to prove the practical feasibility of limit cycle walking.After repeated experiments,the limit cycle walking without power was successfully realized in both the virtual prototype and the actual physical prototype.In the experiments of virtual prototype and actual physical prototype,we found that because the actual model is difficult to search for suitable initial conditions with the help of computer search,it is necessary to rely on a large number of manual experiments to test and find suitable initial values to realize the limit cycle walking.Walking itself has more stringent requirements on the walking environment.In order to improve these shortcomings of pure passive limit cycle walking,this paper adds a central pattern generator as a drive control based on the original model to provide active torque drive for the passive biped model,and uses genetic algorithms to search and optimize the central pattern generator Parameters.The results show that the passive biped robot based on the control of the central pattern generator can realize the limit cycle walking more easily,and has better adaptability to the slope of the inclined plane,so as to achieve more stable limit cycle walking.