Research On Yaw Moment Intelligent Control And Energy-Efficiency Optimization For Humanoid Robot

Humanoid robot has drawn a great deal of attention from both scientific and industrial communities over the latest decades,due to its unique bipedal movement and human-like appearance.The problems of locomotion stability and high energy consumption have become the critical difficulties in robotic field and limit the further application of humanoid robot.At present,the control of humanoid robot on stability and energy efficiency is still a challenging problem.The Zero-Moment-Point(ZMP)criterion has been applied in many humanoid robots as one of the most popular locomotion stability criteria,which include ASIMO,HRP and so on.In recent years,research has shown that ZMP criterion can only guarantee the moment balance in the horizontal plane.However,the yaw moment along the support leg is inevitably generated due to the fact that the components of humanoid robot move in different planes during walking,which may result in the imbalance in the vertical plane.Thus,to date,how to improve energy efficiency and keep locomotion stability in the field ofhumanoid robot is still a challenging difficulty.Motivated by the previous works,this dissertation focuses on improving the locomotion stability and energy efficiency of humanoid robots.The mainresearch works are as follows:1.Yaw moment control for humanoid robotutilizing arms swingA novel yaw moment control method based on arms swing is proposed.After analyzing the moment equilibrium conditions of yaw axis,the expression of the moment generated by arms swing in 3D environment is firstly obtained.Secondly,the quadratic programming with inequality constraints is formulated,in which joints accelerationsare minimized to counteract the unexpected yaw moment with energy efficiency.Moreover,an online iteration algorithm based on linear variation inequality(LVI)is adopted to deal with the optimization problem with both inequality and equality constraints and the optimized trajectories of arms swing is achieved.Finally,several simulation experiments are conducted to validate the proposed method.Compared with the prior works,the newly proposed one in this chapter takes the constraints of joint acceleration into account and the problem of peak cut-off caused by the physical constraints of motor is successfully avoided.Simulation results have shown that the yaw moment is effectively eliminated and the energy efficiency is improved with good locomotion stability.2.Yaw Moment control of humanoid robot based on position compensation of ankle jointThe traditional arms-swing-based schemes may not be always applicable in realistic circumstance in the case of handing object-carrying tasks during walking.To overcome these problems mentioned above,an effective framework of yaw moment compensation based on ankle joint control is proposed.To facilitate the analysis,this method replaces the influence caused by the rest part of humanoid robot above the ankle of the support foot with the force and moment.To achieve excellent walking performance,the walking control problem is formulated as a quadratic programming problem by minimizing the accelerations of ankle joints.In order to deal with the complicated optimization problem with inequality and equality constraints,a novel online iteration method is proposed for finding the optimal ankle torque and generating the ideal trajectories of ankle joints.The main advantage of this method is that the undesired yaw moment can be compensated by controlling ankle joint,which can effectively reduce the impact on the original gaits.Finally,the simulation results proved the effectiveness of the proposed method.3.Walking control for humanoid robot using interval type-2 weighted Fuzzy Logic Systems(FLSs)To improve the locomotion stability and energy efficiency,a new walking control scheme based on interval type-2 weighted FLSs is proposed.This presented control system consists of two parts including offline learning and online control.To improve the learning performance,the interval type-2 FLSs is employed to deduce a specific weight for each sample according to ZMP margin,yaw moment and energy efficiency.After that,a weighted extreme learning machine is newly designed.During the phrase of online control,the optimal correction values of all robot joints are achieved for ensuring stable bipedal walking in accordance with ZMP error and yaw moment.Furthermore,the comparisons with traditional ZMP method and SVM method are carried out.The experiment results are provided to verify the good performance of the proposed control scheme.4.Tracking control for humanoid robot with input nonlinear constraintsTo compensate the backlash effect which exists in robotic system,an adaptive tracking controller is proposed,which takes input nonlinear constraints into account.In the controller design,a novel backlash inverse function is incorporated to cancel the lines-segments effect.Compared with traditional controllers,the proposed one can compensate the unknown actuator backlash nonlinearity with online parameter estimations without the prior knowledge of backlash parameters.Furthermore,the effectiveness and robustness of the proposed controller have been examined.5.The development of humanoid robot and walking experimentsA novel design scheme of humanoid robot walking system is proposed,which contains two parts including host computer control system and humanoid robot.The host computer is equipped with high-definition camera,by which the positions of robot and target object are recognized in real time.After collecting position information,the robot can be remotely controlled through wireless network.To improve the robustness of walking control system,the architecture of master-slave dual CPUs is adopted,which has the advantages of good scalability and robustness.Experiments show that the humanoid robot,which is designed with the proposed solution,can complete the walking well.