Research On Brain-Controlled Mobile Robot Combining Brain-Computer Interface And Model Predictive Controller

At present,brain-computer interface has broad applicatio n prospects in many fields,among which brain-controlled mobile robot is an important research field of brain-computer interface.Mobile robot based on brain-computer interface control can provide safeguard of effective life for disabled people,which have very high scientific research value.This paper mainly studies how to combine the brain-computer interface with the controller to realize the motion control of the mobile robot by the brain’s intention,and improve the safety performance and control performance of the brain-controlled system.First,formulate the research plan of brain-controlled mobile robot.The brain-controlled mobile robot system includes five parts: operator,brain-computer interface model,controller,sensor and mobile robot.In the brain-computer interface model,the conversion strategy between brain control instructions and control instructions is determined.In order to improve the security,real-time and continuity of the motion control of the brain-controlled mobile robot,a shared control strategy is introduced,and the brain-computer interface is combined with the model predictive controller to further optimize the control instructions.The environment perception of the brain-controlled mobile robot is realized by lidar.Secondly,the feature extraction and classification algorithms of different EEG(Electroencephalography)signals are studied.In the process of studying the EEG signal of motor imagery,a multi-dimensional feature classifier of EEG signal is proposed,which further improves the recognition accuracy of multi-classification of motor imagery.In the research on the classification of EEG signals of SSVEP(Steady-State Visual Evoked Potential),the ECCA classifier is used to classify SSVEP signals,and a high recognition accuracy is achieved.Comparing the classification results of the two EEG signals,the EEG signal of SSVEP has a shorter sampling time and higher recognition accuracy under the ECCA classifier.Therefore,the EEG signal of SSVEP is selected as the control signal of the brain-controlled mobile robot system,using the ECCA algorithm as its classifier.Thirdly,the model predictive controller of brain-controlled mobile robot is studied.The prediction model of the model predictive controller uses the kinematic model of the mobile robot,and derives the discrete state space equation according to the kinematic model,so that it can predict the future state of the mobile robot.The cost function and constraints of the controller are deduced,so that it has the abi lity to track the intention of brain control and avoid obstacles,and complete the construction of the mathematical model of the controller.The controller is simulated offline in the ROS system to verify the effectiveness of the controller.Finally,the online simulation experiment of the brain-controlled mobile robot was carried out,the experimental plan was designed,the experimental platform and experimental scene were built by using the brain interface device and the ROS operating system,and 4 subjects were invited to participate in the online simulation experiment,which was task of the brain-controlled mobile robot.By analyzing the completion of the brain-controlled tasks,it is verified that the brain-controlled mobile robot can control the mobile robot in real time through the EEG signal under the optimization of the model predictive controller,without any collision,and has good safety performance and control performance.