Research On Trotting Locomotion Control And Center-of-Mass Correction Of High-performance Quadruped Robots

People are inspired by legged animals in the natural world a lot because of their perfect flexibility of locomotion and adaptability to rugged terrains.In recent decades,researchers around the world have been focusing on varies fields of bionic legged robots,including the bionic structure,the locomotion mechanism,the driving style,the energy efficiency,and the sensing technology.Legged robots are expected to accomplish tasks such as cargo transportation,engineering operation,exploration and rescue in unstructured environments,to make up for the shortages of wheeled and tracked robots.At present,the topological structure of legged robots includes biped,quadruped and hexapod.Among them,quadruped robots become major research objects because they have good performance both in stability,flexibility,and efficiency.The Trot gait is widely used by quadruped robots because it has good dynamic response,stability and symmetry properties.This thesis focuses on the bionic quadruped robots developed by our research team.Firstly,the onboard control system and the human-robot interaction system are designed and integrated based on the requirments of the current prototype platforms.The basic motion control is implemented based on the modeling of the leg.Then,the locomotion control method based on the foot trajectory planning is proposed and the robustness of quadruped robots is improved.The optimal force distribution method for the supporting legs of quadruped robots in Trot gait motion is proposed and the adaptability to rough terrains of quadruped robots is improved.Furthermore,the method of online center of mass(CoM)estimation and motion adjustments for quadruped robot suffering payload position changes is proposed.The thesis mainly includes the following research contents.1.Aiming at the control and human-robot interaction requirments of the current hydraulically actuated quandruped robots,the onboard control system is designed and integrated.A multimode human-robot interaction system which has control,debugging and monitoring functions is developed.The works above provide the hardware support for the realization of locomotion control.The topological structures of the leg and the joint are analyzed,and the driven model of joints and kinematic/dynamic model of legs are built.For the motion control of legs,the impedance control method based on the servo of joint positions and the virtual model control method based on the servo of joint torques are proposed.Experiments of leg motion control are implemented on the prototype of SCalf,which provides the foundation of locomotion control of the robot.2.Aiming at the problems that the quadruped robot lacks flexibility and dynamical response to disturbances,the stable locomotion control method in Trot gait motion based on the inverted pendulum model is proposed inspired by the virtual leg theory.The mathematic relation between the body attitude stability and the speed of the supporting foot is established.The qualitative relationship between the velocity stability and the landing position of the swing leg is established.The omnidirectional stable locomotion control of quadruped robots on varies terrains is implemented based on the planning of foot trajectory and compliant motion control of legs.This method is easy to be implemented on hydraulic platforms and the ability against disturbances of quadruped robots is effectively improved.3.Aiming at the problems that the quadruped robot suffers violent foot-ground interaction impacts and lacks adaptability to rugged terrains,the locomotion control method based on the model predictive control is proposed.The driven model of contact forces on system states is analyzed and built.The prediction of the state space is realized based on the derivation of the dynamic equation in the iterative form.The force distribution is transformed into a quadratic optimization problem,realizing the optimal tracking of the state trajectory with the lowest consumption of energy.The adaptation strategy to rugged terrains is proposed based on the concept of pseudo attitude angle.Simulation and experiment results verify that the adaptability to rugged terrains,the robustness and the energy efficiency of quadruped robots is improved with the proposed method4.Aiming at the problems that the center of mass(CoM)position is difficult to be estimated in real time for quandruped robots in Trot gait motion,and the CoM changes have influence on locomotion performance of the robot,the online CoM position error correction method is propsed.The underactuated characteristics of the quadruped robot and the influence mechanism of the CoM modeling error on the locomotion performance.The qualitative relationship between the CoM error and the locomotion state is derived.The online CoM estimation and motion adjustment method is poposed based the state feedbacks.The locomotion adaptation of quadruped robot to payload position changes is realized.Simulation and experiment results verify that the adaptability of quadruped robots to payload disturbances is improved.