Research On EEG Control Method Of Mobile Robot In Unstructured Environment

For robots,most of the environments in human life are unstructured.Due to the uncertainty of environmental information,many structured environments tend to become unstructured environments in human daily life.Therefore,studying the control method of mobile robot in unstructured environment has great practical value and profound practical significance.In traditional robotic systems,human-computer interaction is often based on physical hardware,such as keyboard,mouse,joystick,touch screen and so on.However,these entity controllers require operators to exercise a certain degree of mobility,which is not applicable to disabled or elderly people with motor dysfunction.Therefore,brain computer interface(BCI)comes into being.Through the BCI,some people who lose their ability to exercise can also control the robot systems as normal people do.It is of important theoretical value and practical significance to study BCI technologies in the control of nonholonomic mobile robot system,especially in the case that robotic devices can be controlled directly through the electroencephalograph(EEG).Aiming at restoring the sport ability of the disabled to a certain extent,this paper focuses on BCIs that can meet the needs of the disabled and develops the corresponding software.The research platforms of this paper are respectively a traditional wheeled mobile robot,a mobile dual manipulator and a smart wheeled chair.Major research content of this paper is summarized as follows.(1)For wheeled mobile robot system:(i)Under the constrain of unstructured environment,an APF-based control law of the brain signals is put forward and the relationship between the strength of EEG signals and the intensity of the APF is built up so the change of EEG signals reflects the variety of the resultant potential field.(ii)A multi-function brain-robot interface(BRI)embedding a SLAM function that acquires information of the odometer,the vanishing point and the door plates to accurately locate the robot,is proposed in the unknown corridor environment.(iii)Image-based visual feedback control system,which can transmit information of the environment to the BRI and implement the teleoperation in real-time,is integrated into the steady state visual evoked potentials(SSVEP)based brain-actuated system.(2)For mobile dual-arm robot system:(i)A motor imagery based teleoperation strategy of a mobile dual-arm robot system performing manipulation tasks in an unstructured environment is proposed.The proposed CSP-SVM recognition algorithm would transform neural activity to control signals for the robot.(ii)A primal dual neural network(PDNN)based redundancy resolution at velocity-leve in the unstructured task space is designed to control the redundant biomimetic robotic arms.The neural dynamics optimization method can avoid decomposing the robot kinematics into task subsystem and posture subsystem,and avoid the various physical and environmental constraints.(iii)Image-based visual feedback control system,which can transmit information of the task space to the BCI and implement the intention-teleoperation of operators in real-time,is developed.(3)For smart wheeled chair system:(i)An outdoor navigation control strategy in an unstructured environment is proposed,which adopts the CCA-classified EEG signals as system input.(ii)A localization method for outdoor navigation is explored using data from odometer and satellite.And an outdoor environmental mapping technique integrated with road edge detection technology,is proposed.