Design And Motion Control Of An Omni-directional Mobile Robot

The most popular wheeled mobile robots are equipped with two independent driving wheels. Since these robots possess 2 degrees-of-freedom (DOF), they can rotate about any point, but can not perform holonomic motion including sideways motion. To overcome this type of motion limitation, omni-directional mobile robot (ODMR) was proposed. They can move in an arbitrary direction without changing the direction of the wheels, because they can achieve 3 DOF motion on a 2-dimensional plane. This capability provides them with excellent mobility, especially in areas congested with static or dynamic obstacles, such as robot soccer field, workshops, warehouses and hospitals. Therefore the ODMR attract much interest among researchers, and they are widely used in the robot soccer.This thesis builds the kinematics and dynamics model of three and four wheels ODMR. Based on the kinematics model, the maximum velocity and acceleration of ODMR in an arbitrary direction is analyzed. Besides, the relation between control sensitivity and mechanical parameters of ODMR is studied by using matrix theory. According to those analyses, the robot is designed as three wheels ODMR with angle of 120 deg between wheels.The control system of ODMR uses two-layer structure. Upper layer is decision-making and communication layer which consists of ARM9 embedded development board SBC2410X. Under-layer is motor control and drive layer, and it consists of three pieces of LM629 which is a dedicated motion-control processor. Those two layers are connected by the parallel port of SBC2410X which responsible for finishing the read-write of control parameters and commands, thereby achieving the accuracy control of the driving motor. The main control program is developed on ARM Developer Suite (ADS, 1.2version), and online debugged through H-JTAG. This thesis also studies the time-optimal control of ODMR, and proposes a practical optimization algorithm by using the maximum principle. Because unreasonably assume the DC motor as a proportional component in this algorithm, this time-optimal control greatly reduce the path tracking accuracy. In order to overcome this disadvantage, DC motor dynamic model is built, which replace the unreasonable assumption. New method greatly improves the path tracking accuracy. Finally, the rationality of hardware and the validities of some theory proposed in this thesis are tested.