Research On Inverse Kinematics And Sliding Mode Variable Structure Trajectory Tracking Control Of Mobile Manipulator

A mobile manipulator consists of a mobile base and a manipulator arm mounted on the mobile base. It is the combination of a mobile robot and a manipulator which substantially increases the performance capabilities of the system, such as much larger workspace than the fixed-based manipulator, and kinematics redundancy.Recognizing these advantages,mobile manipulator is used in numerous application areas. For instance,these systems have been suggested for tasks involving hazardous environments, such as waste management and space operations, and for application in the manufacturing sector. But the combination of the mobile robot and manipulator also increases the complexity of the system which makes controlling and motion planning of the system difficult. For example, mobile manipulators of this type possess a complex and highly coupled dynamic models. Additionally, obtaining base mobility by using a wheeled mobile base results in the introduction of nonholonomic constraints on the system kinematics, and it is well known that the traditional control methods are insufficient for nonholonomic systems.Thus coordinated control of mobile manipulator systems is challengeable in both theory and application.In this paper, the control strategy and ways of mobile manipulator and the history and development of sliding mode control are presented,the tracking control of mobile manipulator is systematically studied.Based on the uniform dynamic model of mobile manipulator, a sliding mode variable structure controller is designed .The main contents are made up by the following parts:1. A new way to build the model of the inverse kinematics of the mobile manipulator is proposed in this paper. Neural network is used to express the nonlinear function between the input and output of the inverse kinematics of the robot. And ant colony optimization(ACO) algorithm is used to learn NN.Artificial neural network (ANN) ,which can express nonlinear system and has strong learning ability, is one of the most popular intelligent controlling ways and it is used popularly. The most important feature of neural network is its learning algorithm which aims to adjust its weights to get a smaller output error. But the traditinal BP algorithm has some defects such as low convergence speed, easily getting a local minima and high requirement for its functions.As a new kind of metheuristic,ACO algorithm is investigated by many researchers.ACO has several characters such as distributed computation, positive feedback of information and heuristic searching. An individual can easily convergence to the local optimum,but many individuals can make problem solution towards global optimum by cooperation. ACO shows a strong ability in hard combinatorial optimization problems (COPs). Continuous optimization is a new research field for ACO. Yet the learning of NN is a typical continuous optimization problem and there are always many parameters needed to be optimized.ACO algorithm is used to learn neural network. It overcomes the shortcomings of traditional BP algorithm. Because ant colony algorithm is mainly used for combination optimization, the basic ant colony algorithm is modified.Two ways for ant colony algorithm to be used to study NN are given:(1) The basic ACO algorithm is extended to continuous domain through adding Local searching and definite searching to the basic ACO algorithm.(2) The basic ACO algorithm is extended to continuous domain through changing the discrete pheromone matrix and probability matrix into continuous function.The new algrithm has the merits of both ACO algorithm and neural network. It also provides a new way for the inverse kinematic modeling of robot through constructing the inverse kinematic model of mobile manipulator by ant colony-nerual network and overcomes the shortcomings of traditional algorithm.2. Newton-Euler approach is used to derive the dynamic model of wheeled mobile manipulator.Since weeled mobile manipulator is subject to nonholonomic constraints, the angles of the two wheels and arms are taken as the generalized coordinates,Newton-Euler dynamic equation is used for every sub-component of weeled mobile manipulator to derive the whole dynamic model of it. The Newton equation of every sub-component is derived in the absolute coordinate system. The Euler equation of each sub-component is derived in the corresponding coordinate system fixed with it.3 . A sliding mode variable structure controller is devised for mobile manipulator.Sliding mode control has strong robustness. When a system is in a sliding mode, it is insensitive to parameter variations and disturbance. Precise dynamic models are not required and the control algorithms are easy for implementation. All these properties make the sliding mode control an ideal candidate for robotic control. But , sliding mode control also has shortcomings.On one side,the traditional sliding mode control has a remarkable drawback that is chattering, which almost makes it impossible in practical application.On the other side, the convergence rate is also an important factor which affects its performance .So how to impare the chattering and improve the convergence rate are two problems to be soluted for the sliding mode variable structure control.Aiming at improving the convergence rate of sliding mode variable structure control, a new nonlinear sliding mode surface is proposed. When the system reaches any point of the new sliding mode surface, it can converge to the equilibrium point with a higher speed than both linear sliding mode surface and terminal sliding mode surface which means the system can get a high convergence speed in the sliding phase with this new sliding mode surface.At the same time, a new two-power reaching law is proposed which makes the system have a higher speed than exponential and power reaching law in the reaching phase.With this new reaching law, the chattering is also eliminated which often happens in traditional sliding mode variable structure control.4. A dynamic sliding mode variable structure controller is devised for mobile manipulator based on the proposed new sliding mode surface.In conventional sliding mode control,the sliding mode surface only relies on the systm state, so the control input is discontinuous.As a consequence of this, chattering is generated.In dynamic sliding mode control,a more general classes of sliding mode surface are used which also include the presence of inputs and, possibly, their time derivatives.Most undesirable effects of discontinuity are transferred to the derivatives of the controlled variables, thus a smoother controlled signal can be get and the undesirable effects of chattering are effectively eliminated .At the same time,the new sliding mode surface and exponential reaching law are used to improve the convergence rate.