Studies On Control Method Of Wheeled Mobile Dual-arm Robots Based On BCI

Brain-control technology presents huge potential in areas like medical rehabilitation,military science,scientific research education,home automation,and entertainment fields.Currently,there are still many problems in instantaneity,stability,accuracy,safety,robustness and adaptability of the brain-control technology of robot based on the Electroencephalograph(EEG)signal.To address the deficiency of current brain-control technology of robot,the study proposes brain-control methods of a kind of compound robot,namely,the wheeled mobile dual-arm robot.Based on the combination of the brain-computer interface(BCI)technologies and the robot control strategies,this study investigates solutions for the asymmetric manipulation problem of the brain-controlled redundant mechanical dual-arms under physical constraints,problems of the navigation and obstacle avoidance of brain-controlled nonholonomic wheeled mobile platform in complex environments,the problem of concurrent manipulation of brain-controlled multiple nonholonomic wheeled mobile platforms under physical constraints.Additionally,theoretical analysis and experimental verification are also performed.Hence,the high-efficiency humanrobot interaction of wheeled mobile dual-arm robot is achieved.The key points of this study are summarized as follows:i)Through analyzing the physiological characteristics of SSVEP signals,a multi-channel collecting,multi-step filtering and multi-algorithm decoding SSVEP-BCI system was designed.In addition,the technical realization process of EEG acquisition conversion,filtering processing,feature extraction,and classification recognition were also clarified.The experiments verified the performance of the system,and compared the recognition effects of the four EEG recognition algorithms,respectively,power spectral density analysis(PSDA),support vector machine(SVM),canonical correlation analysis(CCA)and multi-variable synchronization index(MSI).ii)For the asymmetric manipulation problem of the brain-controlled redundant mechanical dual-arms under physical constraints,a brain-control strategy is proposed to control the relative motion of the end effectors of the arms.The relative Jacobian matrix is firstly applied to simplify the two-arm motion planning problem into a single-arm motion planning problem;Further,a polar coordinate brain-control method,which is more safe than the conventional Manhattan grid brain-control method,is proposed to control the movement of the end effector of the simplified mechanical arm in a defined plane.The EEG classification results correspond to the steering or radial movement of the end effector in the polar coordinate system to plan its trajectory in the defined plane.Due to the redundancy of the robot arm,there are multiple or infinite solutions when the trajectory of the task space is transformed into the joint space.To find the optimal solution under the optimization criterion with the consideration of physical constraints such as joint position and joint velocity constraints,this motion optimization problem with constraints can be transformed into a quadratic programming(QP)problem.Finally,the primal-dual neural network(PDNN)based on linear variational inequalities(LVI)is used to reduce the computational burden and solve the QP problem to obtain the optimal joint angular velocity online.Therefore,the optimal trajectory of the joint space of the mechanical arm can be achieved,the joint deviation angle would be avoided;and the repeatability,instantaneity along with safety of the relative movement of the brain-controlled mechanical dual-arms can be guaranteed.iii)For the navigation and control problems of brain-controlled nonholonomic wheeled mobile platform in complex environments,a semi-autonomous brain-control navigation and obstacle avoidance strategy is proposed for the environment of slippery ground,disorderly colors,uneven light,and multiple obstacles.Initially,a simultaneous localization and mapping(SLAM)method based on the multi-information fusion is proposed to ensure the real-time positioning and precise mapping of the robot in the condition of ground sliding,disorderly background color and uneven lighting.Further,an artificial potential field(APF)with global position and posture convergence properties is proposed for the safe path planning without collision in the obstacle environment,which handles nonholonomic constraints to achieve the convergence of robot's position and posture while avoiding obstacles.In order to adapt to the ground slip,a differential flatness-based robust controller is used to track the planned instantaneous trajectory,which comprehensively considers the lateral,longitudinal and steering slip to effectively offset and restrain the sliding disturbance.The navigation scheme combined by the above three achieves the closed-loop structure of positioning update,path search,trajectory generation and trajectory tracking in each control cycle of the nonholonomic wheeled mobile platform,realizing the accurate,stable and efficient online navigation.Finally,based on the above navigation strategy,a brain-control method for wheeled mobile robot is creatively presented to alter the distribution of artificial potential fields via the EEG signal.The conversion relationship between EEG classification results and the distribution of the APF acting on the nonholonomic wheeled mobile platform is defined,which builds the intuitive mapping between the human control intention and the robotic obstacle avoidance behavior.Under the influence of APF,the robot will automatically converge to the target position and posture.This semi-autonomous brain-control method can effectively reduce the workload of the operator.iv)Regarding the issue of concurrent manipulation of brain-controlled multiple nonholonomic wheeled mobile platforms under physical constraints,A leader-follower formation control method based on nonlinear model predictive control(NMPC)is proposed,and the braincontrol operation of multi-robots is achieved by brain-controlling the piloting robot in the formation system.Firstly,two feature points are set up on the brain-controlled piloting robot.According to the camera perspective projection model,the vision-based formation kinematic model for multiple nonholonomic wheeled mobile robots is established by the mapping relationship between the image coordinate system,the camera coordinate system,the following robot coordinate system and inertial coordinate system.Thereafter,the model predictive control is used to stabilize the system while considering the nonholonomic constraints,input constraints and state constraints of the system.The problem of minimizing the cost function in the receding optimization process of model predictive control is transformed into a QP problem;Finally,the optimal solution for the QP problem of the formation control can be quickly solved online by LVI-PDNN.By means of the proposed leader-follower formation control method,the following robots can be stabilized from different initial states to the state of motion led by the braincontrolled piloting robot.Because the model predictive control is a receding horizon closedloop optimal control strategy with good anti-interference ability.Moreover,various physical constraints in the formation system are fully considered in the control.In addition,the optimization problems in the model predictive control can be quickly solved by the primal-dual neural network in real time.Therefore,the proposed formation control method correspondingly guarantees stability,robustness,smoothness,safety,and instantaneity of the concurrent manipulation of the brain-controlled multi-robots.