Stability Analysis And Coordination Control Research Of Biped Robot For Running

For the biped robot's potential application, related researches have aroused much attention since its appearance. Due to its unique motion pattern, biped locomotion is the most fundamental and challenging problem undoubtedly. And biped walking is always the focus of this field and remarkable achievements have been acquired. But biped running has been becoming a research hotspot until recent years and only a few achievements can be seen. The running speed of the existing robots developed by some institutes is still very low. And the running action of them is not flexible enough. Therefore, it achieves far from the expectation of human being. The main reason is that some theories related to biped running are still not further studied besides the technological constraints on the chips and actuators. So some related theories brought by biped running are studied in this dissertation. They mainly include mathematical model of biped robot for running, the stability of biped running, running gait planning and coordination control strategy. The detailed descriptions are as follows:Firstly, as the basis of research on biped robot for running, the hybrid switched system model is built. Several models and equations are founded respectively. And they are kinematics model of biped robot for running by using modified D-H reference coordinates and rooted-system configuration, dynamics equations during the stance phase and flight phase by using Lagrange function, states-switching models between phases by using switching conditions. Then, the whole mathematical model of biped running is built using the hybrid switched system with eight tuple.Secondly, the stability of biped robot for running is studied based on analyzing the existing stability criterions of biped walking and running. Then, the stability description and intuitive expression for biped running are developed. And the stability definition for biped running is proposed from the view of avoiding falling down. The general stability criterion for the whole biped running process are also proposed, which combines the lost balance degree criterion in the stance phase with the angular momentum criterion in the flight phase.Thirdly, the biped running gait planning is discussed. As the reference and basis for the running gait planning, the biomechanics law of human being running is studied. Then, the biped running gait planning is formalized as the constrained optimization problem. The Genetic Algorithm is used to choose and optimize the gait parameters. The minimum torque and energy are chosen as the cost function of Genetic Algorithm respectively. And the strategies combining rejection with penalization are used to deal with all kinds of constraints. The simulation and contrast of biped running gait are carried out based on these two cost functions. The result of simulation shows that the running gait planning can be realized by using Genetic Algorithm and it is feasible and effective.Fourthly, the control strategy of biped robot for running is investigated. And the time-invariant control strategy with feedback based on dynamics model is proposed. The dynamics model is formalized as the nonlinear affine system. Then, the original system is transformed into double integrators by using feedback linearization. The finite-time stability controller is designed for the linearized system. Meanwhile, the dynamics singularity is analyzed which results from the noninverse of decoupled matrix during the process of linearization. Based on the control strategy discussed above, two cases for research are given. One is the planar, five-link, biped robot for running without feet. The other is the planar, seven-link, biped robot for running with feet. For the former case, the time-invariant feedback control laws during the stance phase and the flight phase of running are both designed based on virtual constraint. And the simulation of the former control law with the ideal model are performed under the two cases in which the running speed are 1.5m/s and 2.5m/s respectively. Then, the partial robustness about the control law is analyzed, which considers flexible impact model and model parameter error. For the latter case, the output function and control law are designed based on zeroing output. And the main parameters in the output function are adjusted online, which have an important influence on running speed and stride. Then, the simulation of the latter control law is implemented for the case in which the biped robot runs on the even ground. In conclusion, the simulation result shows that the time-invariant control strategies with feedback based on dynamics model have strong robustness and good stability. And the comparison of running control strategies between the two planar biped robots is discussed, including the same and the difference aspects.Finally, the main contents of this dissertation are summarized and key points for future researches are discussed.