Research On The Dynamic Analysis And The Trajectory Planning Of A Spherical Robot

Spherical robots are a new type of intelligent robots with a global shell, and take the rolling as their main motion mode. The spherical robots have the advantages such as: obtaining the great steady ability after being imposed a disturbance, achieving the omnidirectional rolling and agile swerving compared with wheeled robots and tracked robots and humanoid robots, because of the special shell and motion mode. Spherical robots are still studied in laboratories at present. A number of spherical robots with different driving mechanism come forth in these years. But few peoples have completed the dynamic analysis and study of the spherical robots. The research on the motion performance analysis and control of the spherical robots can not be developed well for this reason. So this work mainly studies the dynamic analysis and the motion planning of a kind of omnidirectional rolling spherical robot with a stable platform.At the beginning, the general situation and the intending tendency of the research on the spherical robots is stated. The existing main problems in the research on the spherical robots follow. And then a kind of omnidirectional spherical rolling robot with a stable platform is introduced. This spherical robot is driven by four linear stepper motors on four screwed spokes in the shell. The whole system has geometrical symmetry in the three dimensional space. A motion planning method is presented based on the structure and the motion model of the robot and the motors'discrete motion nature. This motion planning method can drive the spherical robot to the destination by finding the most appropriate combinations of the four motors'motion states at each step of the spherical robot motion process. The simulation of ADAMS testifies the motion planning method. To study the motion performance and the dynamic control of the spherical robot further, the complete dynamic model of the robot on a plane and an inclined plane are established respectively using the Lagrange-routh equations. The strategy and the approach of how to eliminate the Lagrangian multipliers that were brought because of the spherical robot's nonholonomical constrain are given. The dynamics analysis and simulation of some typical examples are developed, and the results validate the effectiveness of the dynamic model and the method. Based on the diagonal recurrent neural networks PID control, the controller of the spherical robot is given after getting the complete dynamic model of the spherical robot, using the modern control theory and the knowledge of the neural networks. The spherical robot can reach any arbitrary appointed destination by controlling the inputs of the four stepper motors. For a number of missions, it is shown that the method agrees well with the results.