Adaptive Gait Planning And Control Of Quadruped Robot Based On Underactuated Configuration Description

With the increasing demand for deep space exploration,the use of mobile robots to land on exoplanets has become an essential step.Current exoplanet landing equipment such as lunar rovers and Mars rovers are all wheeled robots,which are not capable of handling gullies,fractures and narrow terrain.Bionic quadruped robot,with its flexibility and stability,can be used not only for outer planet exploration,but also for scientific research and emergency rescue in extreme environments.In addition to the performance of traditional robots used on Earth,robots need to face multiple challenges such as lack of sensory information,reduced human controllability,and inability to repeat training in advance.In order to improve the adaptability and anti-interference of the robot controller to unpredictable environments,this paper combines model predictive control to design an adaptive gait planning controller for quadruped robots,which improves the adaptive capability and stability of the robot's motion to unknown external environmental factors.The control efficiency of unstable support configuration directly affects the stability of motion,which is affected by external environment during robot dynamic motion.By analyzing the typical mechanism design methods and gait characteristics of quadruped robot,the necessity of control efficiency analysis of unstable support configurations is introduced.By analyzing the kinematics and dynamics of single and double legged robots,dynamics equations are established based on equivalent leg theory and Lagrangian function.Based on leg's Jacobi matrix and principle of virtual work,the relationships between variables and forces in task space,joint space,and equivalent leg space are derived,and the equations for constraint of plantar friction cone represented by the equivalent leg variables are obtained.By decomposing the input matrix,the underactuated characteristics of single and double legs support systems are demonstrated,and the controller is transformed into an underactuated system control problem and a force distribution problem with constraint optimization.The underactuated controller is established based on local feedback linearization,and the control singularity of single and double legs support phases is proved by the selection method of actuated variables.The tracking effect of controller on selected actuated variables is verified by dynamics simulations,which is used to guide the design of optimal force distribution method for the quadruped robots.When a robot performs exploration tasks independently,it encounters unexpected conditions that lead to abrupt changes in control commands,which is a great challenge to motion stability;meanwhile,unpredictable complex situations such as rugged terrain,external impulse,and eccentric loads can disrupt the robot's gait cycle and affect the base control accuracy and task execution capability.The selection of foothold for point-footed robot determines its ability to adapt to complex terrain,external impulse and other harsh environments.Based on the finite state machine,a system state monitor is designed for state variable synchronization and swing phase switching among the modules,and thus a time-event hybrid triggered quadruped robot control system is established,which optimizes the transition link of the base motion command and the foothold selection method,and constitutes a quadruped robot adaptive motion planning framework.Simulation results show that the local complementary optimal force distribution approach improves the robot's adaptability to control parameters and gait.Mobile robots often pre-plan their walking trajectories through visual navigation in practical applications,and the gait pattern shows a certain periodicity.The simulation results show that the controller reduces the impact effect of touching ground on base motion compared with the traditional design method,and the increase of prediction time increases the smoothness of base motion while prolonging the single-step simulation time.Finally,dynamics simulation tests and comparative analysis were conducted for several typical application scenarios.From the simulation results,it can be seen that the hybrid trigger gait planning controller based on the model predictive control algorithm and the local complementary optimal force distribution method effectively improves the robot's adaptability to unpredictable irregular ground,sloping terrain,lateral impact and large mass eccentric load,which makes the robot more adaptable and stable in the face of unpredictable and complex environments.By studying the adaptive planning and control algorithms of quadruped robots,we can effectively enhance their ability to autonomously handle unpredictable environments or disturbances during the execution of tasks and improve the stability of movement on different terrains,which is of great significance for the more mature application of robots in places such as outer planet exploration,military reconnaissance,scientific investigation and rescue in extreme environments.