Research And Realization On The Autonomously Walking Of Humanoid Robot On Uneven Pavement

In recent decades,with the rapid development of robot technology and the growth of robots's level in intelligence,the robot is from the production area,and gradually to the development of life,and the robot has a broad application prospect in life.Realizing the autonomous walking on uneven pavement is the precondition for the robot to finish other tasks.Autonomous walking is mainly reflected in the "autonomy",followed by "walking".It requires the robot can identify the environment and stably walk in the environment,and have strong adaptability to the environment.Different from the traditional way of teaching and playback,in thesis,the stair structure pavement was chose as the environment,by modeling the motion model of the robot walking and study the autonomous walking of humanoid robot on uneven pavement combine the vision with gait planning.First of all,this thesis introduces the methods of feature extraction and the method of stair feature detection.The imaging model of the camera and the transformation relationship between each coordinate system are introduced.The method of single camera calibration is studied.According to the motion of robot,the transformation relation between image coordinate system to camera coordinate system and camera coordinate system to the world coordinate system is presented.Secondly,based on the robot kinematics model,introducing the inverted pendulum model plan when the robot is walking centroid and ankle trajectory,while adding the tilt angle to improve the stability of the climbing of the robot in the process of using the inverse kinematics analysis method to calculate the trajectory for each joint angle sequence.Finally,the improved genetic algorithm is used to optimize the gait parameters,and the optimal solution is obtained.The parameters are used to realize the autonomous walking of the robot on uneven pavement.