Adaptive Control Method Research Of Multi-modal Intelligent Wheelchair Based On Nonlinear And Uncertainty Observer

Intelligent wheelchair has become an effective travel tool for the lower limb disabled and the aged.In fact,it is a special form of the mobile robot.The main difference between the intelligent wheelchair and the ordinary mobile robot is that the user played a dominant role in control and operation process,while the common mobile robot lies in totally automomous control.Thus,the various control algorithms and control modes must be based on user's subjective consciousness in real-time dynamic control.The system fully refelccts the structural characteristic,which is human-in-the-loop.With the increasing demand for the quality of daily life and social activities for the disabled and the aged,the research on the real-time and safety for the intelligent wheelchair control has become the focus of researchers.The main problems of the current widespread wheelchair include the following aspects.Firstly,there are deficiencies on the control mode,because most users are unable to choose the proper control mode of human-computer interaction according to their own characteristics.It is difficult to satisfy the disabled users' requirements for different degree of disability.Secondly,in some narrow indoor environments,the conventional wheelchair with differential steering system exits limitations on movement form.It has large turning radius,and the system safety can not be guaranteed,which might make the wheelchair tilt and cause some security matters.Thirdly,in a large number of environmental data processing,the conventional mobile robot control algorithm is often with high computational complexity,so it can not guarantee the real-time operation.Finally,in the control system of intelligent wheelchair,due to the user's individual differences and the changing environmental parameters,if the nonlinear and uncertain factors in the system are not considered fully,it will inevitably affect the control accuracy and safety of the intelligent wheelchair.In view of the above problems,a novel manual control method based on user's two-dimensional gravity body posture is proposed.As the main control mode of an omni-directional intelligent wheelchair,it is easy to control the intelligent wheelchair for users.Considering real distabances,a real-time autonomous obstacle avoidance,and a trajectory tracking control algorithms are presented,together with the controller design method based on an improved RBF neural network and a nonlinear uncertain observer of an intelligent wheelchair system.The main work of this paper can be summarized as follows:(1)Based on Self-organizing Feature Map(SOFM)neural network,a novel clustering algorithm for two-dimensional gravity center of dynamic human sitting posture is presented.Firstly,the algorithm is optimized and improved in terms of initial weight determination and neighborhood selection of the SOFM neural network.Here,the error adjustment algorithm is based on the neighborhood range,which effectively reduces the complexity of clustering algorithm,and the improved neighborhood selection method increases the clustering convergence speed.The improved algorithm also ensures the real-time judgment of user's motion intention according to two-dimensional gravity center of body posture.At the same time,through weighted correction algorithm for multiplicative habit,this control mode can ensure the security of the system,and a hand-free interactive mode is provided for special users as a flexible control mode selection to control the intelligent wheelchair.Simulation and experimental results show that compared with the conventional SOFM clustering algorithm and other clustering algorithms,the improved SOFM algorithm are significantly improved in real-time and accuracy.(2)On the basis research of intelligent wheelchair obstacle avoidance strategy,Traditional fuzzy obstacle avoidance algorithm in parameter selection mainly depends on the designer's experience,thus an intelligent wheelchair obstacle avoidance algorithm is proposed based on an improved fuzzy neural network.Firstly,an improved fuzzy neural network controller is proposed for obstacle avoidance.Secondly,taking into account the intelligent wheelchair control system's particularity of human-in-the-loop,an improved neural network based on off-line combined with on-line training method is applied to train the fuzzy logic parameters,which reduces the computational complexity of the obstacle avoidance algorithm.Moreover,the experience matrix vectors are introduced to solve the problem of the conflict between the expected motion direction and the computing direction of the obstacle avoidance.Finally,the simulation and experimental results show that the proposed algorithm can effectively improve the real-time performance,intelligence and humanization degree of the autonomous obstacle avoidance of the intelligent wheelchair.(3)In view of the existing user's difference,together with the internal and external environment of bounded unmeasurable disturbances in the intelligent wheelchair control system,an improved Radial Basis Function(RBF)neural network is presented.Firstly,the improved RBF neural network is used to approximate to the nonlinear inverse kinematics model of omnidirectional intelligent wheelchair on line.Secondly,a direct adaptive wheelchair trajectory tracking control method is designed and implemented,and the design method and procedure of the adaptive controller are introduced and given in detail.Furthermore,using Lyapunov stability theory,on condition that the external disturbance exists and the weights of the neural network approximation error is bounded,it is proved in theory that the tracking error is uniformly stable and bounded in trajectory control for the omnidirectional intelligent wheelchair controller.Finally,the simulation and experimental results show that the improved RBF algorithm can effectively approximate the nonlinear inverse motion model of the system,and improve the robustness of the trajectory tracking motion control of the intelligent wheelchair.Meanwhile the robustness of the system is improved.(4)Considering the real-time requirements and system problems inside and outside the high nonlinear and uncertain environmental factors of the intelligent wheelchair,a novel trajectory tracking control method for omnidirectional intelligent wheelchair is presented based on uncertain nonlinear observer.Firstly,the dynamic model of omnidirectional intelligent wheelchair system is established and analyzed in detail.Secondly,the stability of the control system is proved by Lyapunov stability theory.Additionally,using the proposed algorithm,design methods of uncertain nonlinear observer and tracking controller are presented.Experimental and simulation results show that using the proposed method to design the observer can effectively reconstruct and compensate for the unknown parameters and nonlinear system of internal and external uncertain disturbance control errors,and reduce the complexity of control algorithm,to achieve accurate real-time tracking of the intelligent wheelchair.