Studies On Brain-actuated Methods For Nonholonomic Mobile Robots

Nonholonomic constraints are widely found in the wheel mobile robot systems.The physical constraints in mobile robot systems bring challenges to brain computer controlled robots.Studies on the methods of nonholonomic mobile robots system based on EEG signals,by using EEG signals to directly control mobile robots and applying technology of brain computer interface on the nonholonomic mobile robots,have the great theoretical and practical significance.Aiming at the motion control of brain-actuated nonholonomic restraint robots,the method of brain-actuated based on non-vector space is mainly proposed in this dissertation.The contents of brain computer interface based on SSVEP,trajectory planning and motion control of nonholonomic robots are studied.The main contributions of this dissertation are summarized as follows:1.A novel method of brain controlled in non-vector space is proposed.The method is based on Hausdorf distance,mutation equations and non-vector space transformation matrix.The Hausdorff distance is utilized to calculate the differences between two images in order to avoid feature extraction in traditional visual controlled systems.The binocular camera is used to simulate the binocular vision of the human,then the matching errors between the goal image and current image can be obtained in non-vector space.With the obtained errors,the brain can produce motion intention in non-vector space,then the EEG signals by visual evoked can be collected,decoded and the EEG commands in non-vector space will be generated.Since the non-vector space is different from the traditional vector space,the non-vector space commands are converted to commands in vector space through the transformation matrix,and send them to the nonholonomic mobile robots to control their motion in the Cartesian space.2.A brain computer interface system based on SSVEP is proposed,which provides the visual feedback in non-vector space and displays the scene of the environment to achieve human-computer interaction between operator and robot.In order to obtain the EEG commands in non-vector space,two algorithms of AdaBoostSVM and multivariate synchronization index(MSI)are used to decode the EEG signals.The average recognition rates of these two algorithms are 86.95% and 93% respectively.The classification results of these two algorithms are analyzed and compared in this dissertation.3.Considering the physical constraints in the nonholonomic mobile robots,the trajectory planning methods for EEG signals parameterized are designed.Two methods for parameterizing Bezier curve and polar polynomial curve are proposed.Combined with the trajectory planning and the EEG signals classification results,trajectory planning is controlled by EEG signals in real time.For the nonholonomic constraints,the controller based on error model is designed.In this controller,linear velocities and angular velocities are the control variables,and the motion control of robot can be realized by velocities.Utilizing the designed controller,the experiments prove the effectiveness of Bezier and PPC parameterization methods.4.The control methods based on non-vector space are designed for brain controlled single nonholonomic mobile robot and brain controlled mobile robots formation.The teleoperation method using the EEG signals to control single robot and multiple nonholonomic mobile robots.The experiments are done to verify the performance of teleoperation system controlled by EEG signals,and verify that the physical constraints and nonholonomic constraints can be satisfied during the robots motion.5.Considering the constraints of obstacles in limited environment,the artificial potential field is integrated into brain controlled robot system,and the brain controlled method based on artificial potential field is proposed.The function relationship between the intensity of brain signals and distribution of the potential field is established,and the change of the EEG signals will reflect both distribution and potential energy field changes of the obstacles.The change of the potential energy fields is applied to the robot,and the control commands of the robot are generated,and the mobile robot can be navigated and controlled.In the end,the experiments had been carried out to verify brain signals controlling robot in corridor which is a specific small space environment.