3D Modeling And Control Of A Spherical Wheeled Robot

The spherical wheeled robot is a growing area of robotics because of its single contact point with the ground and exceptional agility in motion. However, the dynamically stable features bring flexibility and thin body to the robot which also leads to extremely static instability. Moreover, making dynamically stable robot work stably in complex environment leads to higher performance requirements for drive and control.To deal with the static unstable problem of the robot and improve the adaptability of the robot to the environment, the mechanism and circuit structure are improved in this dissertation. The dynamic models are established and a hybrid motion controller is constructed according to more complex road as well. Moreover, the accuracy of the model and the validity of the controller are tested by simulations and experiments. The detailed research achievement mainly includes:(1) Design and manufacture an annular support leg, calculate the driving requirement and improve the power capacity, propose and design a new type of speed measurement device of the spherical wheel.(2) Based on the improved robot plantform, the constraint conditions and the force are analyzed by using Newtonian mechanics. The kinetic energy of the multi-rigid- body robot is derived by utilizing the kinetic energy equation of particle system and conversion of coordinate methods. On the basis of analytical mechanics principle, the 2D and 3D dynamic models are developed when the robot moves on horizontal plane and ramp way separately.(3) By analyzing the motion of the robot around the equilibrium position, the attitude and speed double closed-loop control structure is constructed using the 2D dynamic model.Through coordinates transforming, the motion in 3D space of robot is transformed into 2D planar motion. Thereby, the control methods used in 2D model is expanded to the 3D model. The fuzzy controller is design for to control the robot when it stands up and climbs the ramp way.The simulations and experiments show that dynamics of the robot is mostly approximated by the 3D model derived in the dissertation. When the robot is unpowered or out of power, it can lean on the ground solidly. When the robot is energetic, the robot can stand up automatically from statically stable state and keep balance. It can also move forward in any direction on the ground, rush to the ramp and climb the ramp if it meets.