SSVEP-based brain-controlled wheeled robot system and continuous sharing control method thereof

Abstract

The invention discloses a brain-controlled wheeled robot system based on SSVEP and a continuous sharing control method thereof, and the system comprises a brain-computer interface module which is used for collecting an electroencephalogram signal generated by a user, decoding the electroencephalogram signal, translating the intention of the user, and obtaining a target control instruction; the smoothing module is used for performing continuous smoothing according to the determined target control instruction to obtain corresponding expected straight speed and angular speed; the sharing control fusion module is used for combining the current state information of the mobile wheeled robot based on the obtained corresponding expected straight-going speed and angular speed and simultaneously considering the surrounding environment state to adjust fusion parameters in real time so as to obtain the sharing speed and the sharing angular speed after fusion iteration is completed; and the dynamic window module is used for calculating a feasible speed window through the external environment information, obtaining a local optimal solution by combining the shared speed and the shared angular speed evaluation component, and controlling the mobile wheeled robot according to the local optimal solution.

Description

Technical Field

[0001] The present invention belongs to the fields of information technology, brain-computer interface, control algorithm, and human-computer interaction, and particularly relates to a brain-controlled wheeled robot system based on SSVEP and its continuous shared control method. Background Art

[0002] Brain-computer interface is a technology that can establish a connection channel between the human brain and external devices. Currently, in the field of biomedicine, it plays an important role in realizing the control of the body by decoding electroencephalogram signals. The brain-computer interface obtains control instructions through direct acquisition of brain signals to achieve the control of external mechanical equipment, creating a brand-new human-computer interaction method. In the field of intelligent robot control, the application of brain-computer interface technology can greatly improve the interactivity and autonomy of robots, enabling robots to perform more flexible and precise actions according to the intentions of operators. The brain-computer interface technology currently has a broad and profound research prospect. On the one hand, it has successfully broken the traditional human-computer interaction method, bringing a more convenient, efficient, and more possible interaction experience to humans, especially special groups with limited mobility. On the other hand, its development and wide application in different disciplinary fields have also opened up more possibilities in fields such as medicine and military.

[0003] The control of wheeled robots is an important field of current brain-computer interface applications. Due to its diversity and flexibility, operators can better select and apply it in different scenarios. By identifying different stimulus signals and imagined signals, etc., the operator can complete the control of the robot, thereby improving the quality of life and social participation of these people.

[0004] Brain-computer shared control is an advanced hybrid control system that combines the decision-making ability of the human brain with the autonomous control algorithm and perception ability of the machine to achieve easier and safer operation. This control method allows users to directly interact with the robot through brain-controlled output. At the same time, the machine makes overall decisions based on the preset algorithm and environmental feedback. The core of shared control lies in complementarity. It not only utilizes the subjective control of the operator to handle complex or uncertain environments but also relies on the precise calculation of the machine to perform precise tasks, thereby improving the flexibility and adaptability of the overall system. It makes the user operation easier by introducing intelligent control, improves the user experience, and improves the execution efficiency while ensuring safety. Of course, there are still quite a few challenges in the current brain-computer shared control, including issues such as continuous control of the brain-computer interface and weight balance of brain-computer control. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a brain-controlled wheeled robot system based on SSVEP and its continuous sharing control method. By collecting EEG signals through the provided new type of stimulation panel, after decoding the EEG signals at the brain-computer interface to obtain the brain control output, the brain control output is smoothed, and the obtained continuous brain control instructions are shared and fused with the current state of the robot's autonomous control. Through multiple iterations, a stable shared output is obtained and applied to the autonomous decision-making dynamic window module of the robot, and the system parameters are kept reasonable according to the current environment during the process, so as to ensure the control accuracy and environmental adaptability of the brain control system.

[0006] On the one hand, to achieve the above object, the present invention provides a brain-controlled wheeled robot system based on SSVEP, including: a brain-computer interface module, a smoothing module, a shared control fusion module, and a dynamic window module;

[0007] The brain-computer interface module is used to collect the EEG signals generated by the user, decode the EEG signals to translate the user's intention, and obtain the target control instruction;

[0008] The smoothing module is used to perform continuous smoothing according to the determined target control instruction to obtain the corresponding expected straight-line speed and angular speed;

[0009] The shared control fusion module is used to combine the obtained corresponding expected straight-line speed and angular speed with the current state information of the mobile wheeled robot, and at the same time consider the surrounding environment state to adjust the fusion parameters in real time, and obtain the shared speed and shared angular speed after the fusion iteration is completed;

[0010] The dynamic window module is used to calculate the feasible speed window through the external environment information, combine the shared speed and shared angular speed evaluation components to obtain the local optimal solution, and control the mobile wheeled robot according to the local optimal solution.

[0011] Optionally, the process of decoding the EEG signals to translate the user's intention includes:

[0012] By decoding the user's EEG signals, the frequency with the maximum correlation coefficient is obtained;

[0013] Normalize the frequency with the maximum correlation coefficient to obtain the proportion probability value of the maximum correlation coefficient in all correlation coefficients;

[0014] Determine the gear values of the speed and angular speed corresponding to the maximum correlation coefficient.

[0015] Optionally, the process of performing continuous smoothing according to the determined target control instruction includes:

[0016] The expected straight-line speed and angular velocity are obtained through the linear combination of the additional speed and additional angular velocity determined by the probability value of the basic speed, basic angular velocity, and maximum correlation coefficient corresponding to the maximum speed and maximum steering angle of the mobile wheeled robot and the gear position in all correlation coefficients.

[0017] Optionally, the fusion parameter is determined by fusing the balance factor determined by the state error generated by user control and the model error generated by the quality of the autonomous control model.

[0018] Optionally, the process of obtaining the local optimal solution includes:

[0019] The speed sampling space is determined according to the maximum speed constraint, sampling time constraint, and safety distance constraint of the mobile wheeled robot. Each set of discretized data in the speed sampling space is evaluated through an evaluation function, and the set of data with the highest evaluation score is determined as the local optimal solution.

[0020] On the other hand, to achieve the above object, the present invention also provides a continuous sharing control method for a brain-controlled wheeled robot system based on SSVEP, including:

[0021] Collect the electroencephalogram (EEG) signals generated by the user, decode the EEG signals to translate the user's intention, and obtain the target control instruction;

[0022] Perform continuous smoothing on the determined target control instruction to obtain the corresponding expected straight-line speed and angular velocity;

[0023] Based on the obtained corresponding expected straight-line speed and angular velocity, combined with the current state information of the mobile wheeled robot, and considering the surrounding environment state in real time to adjust the fusion parameter, the shared speed and shared angular velocity after the fusion iteration are obtained;

[0024] Calculate the feasible speed window through the external environment information, combine the shared speed and shared angular velocity evaluation components to obtain the local optimal solution, and control the mobile wheeled robot according to the local optimal solution.

[0025] Optionally, decoding the EEG signals to translate the user's intention includes:

[0026] Decode the user's EEG signals to obtain the frequency of the maximum correlation coefficient;

[0027] Normalize the frequency of the maximum correlation coefficient to obtain the probability value of the maximum correlation coefficient in all correlation coefficients;

[0028] Determine the gear position values of the speed and angular velocity corresponding to the maximum correlation coefficient.

[0029] Optionally, the continuous smoothing according to the determined target control instruction includes:

[0030] Obtaining corresponding expected straight-line speed and angular velocity through the linear combination of the additional speed and additional angular velocity determined by the proportion probability value of the basic speed, basic angular velocity, and maximum correlation coefficient corresponding to the maximum speed and maximum steering angle and gear of the mobile wheeled robot in all correlation coefficients.

[0031] Optionally, the fusion parameter is determined by fusing the balance factor determined by the state error generated by user control and the model error generated by the superiority and inferiority of the autonomous control model.

[0032] Optionally, obtaining the local optimal solution includes:

[0033] Determining the speed sampling space according to the maximum speed constraint, sampling time constraint, and safety distance constraint of the mobile wheeled robot, evaluating each set of discretized data in the speed sampling space through an evaluation function, and determining the set of data with the highest evaluation score as the local optimal solution.

[0034] Technical effects of the present invention: The present invention discloses a brain-controlled wheeled robot system based on SSVEP and its continuous sharing control method. The concepts of model error and state error are introduced in brain-computer shared control. Through the regulation of the fusion gain, the shared fusion of brain-controlled output and robot autonomous control is realized. By adding a shared speed and angular velocity evaluation component to adjust the evaluation function of the autonomous control model, the burden on the operator's brain is reduced, the fault tolerance rate of brain control is improved, the continuous control of the system is realized, and the control accuracy and environmental adaptability of the brain control system are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation of this application. In the drawings:

[0036] Figure 1 It is a schematic structural diagram of a brain-controlled wheeled robot system based on SSVEP according to an embodiment of the present invention;

[0037] Figure 2 It is a schematic diagram of a new type of stimulation panel for activating brain electrical signals according to an embodiment of the present invention;

[0038] Figure 3 It is a schematic flowchart of a continuous control method of a brain-controlled wheeled robot system based on SSVEP according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0040] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0041] As Figure 1 shown, in this embodiment, a brain-controlled wheeled robot system based on SSVEP is provided, including: a brain-computer interface module, a smoothing module, a shared control fusion module, and a dynamic window module;

[0042] The brain-computer interface module is used to collect the electroencephalogram signals generated by the user, decode and translate the electroencephalogram signals to obtain the user's intention, and obtain the target control instruction;

[0043] The smoothing module is used to perform continuous smoothing according to the determined target control instruction to obtain the corresponding expected straight-line speed and angular speed;

[0044] The shared control fusion module is used to combine the obtained corresponding expected straight-line speed and angular speed with the current state information of the mobile wheeled robot, and at the same time consider the surrounding environment state to adjust the fusion parameters in real time, and obtain the shared speed and shared angular speed after the fusion iteration is completed; wherein the current state information of the mobile wheeled robot includes the two-dimensional horizontal and vertical coordinates, forward speed, steering speed, and heading angle of the mobile wheeled robot;

[0045] The dynamic window module is used to calculate a feasible speed window through the external environment information, combine the shared speed and the shared angular speed evaluation component to obtain a local optimal solution, and control the mobile wheeled robot according to the local optimal solution; wherein the external environment information includes the obstacle information and target point information of the environment.

[0046] Specifically, the electroencephalogram (EEG) signals generated by the user are collected through a new type of stimulation panel. The new type of stimulation panel is a stimulation panel for collecting EEG signals using steady-state visual evoked potentials. The stimulation array in the stimulation panel for activating corresponding brain features is presented in the form of a two-dimensional stimulation surface, and the stimulation surface consists of 21 small strip rectangles in 3 rows and 7 columns. The horizontal dimension of the new type of stimulation panel represents the intention of straight-line speed, including high-speed, medium-speed, and low-speed command signals, and the vertical dimension represents the intention of steering angular velocity, including 7 command signals for left and right directions with low, medium, and fast deflection speeds and no steering. The viewing angle between each small strip is approximately 4 degrees, and the stimulation frequency of the small strips diverges equidistantly from low to high in serial numbers from the inside to the outside, with a range of 6 - 10 Hz and an interval of 0.2 Hz. The visual stimulation phase difference between each adjacent small strip is π / 2.

[0047] Furthermore, the process of decoding and translating the EEG signals to obtain the user's intention includes:

[0048] By decoding the user's EEG signals, the frequency with the maximum correlation coefficient is obtained;

[0049] The frequency with the maximum correlation coefficient is normalized to obtain the proportion probability value ρ of the maximum correlation coefficient among all correlation coefficients. max ,

[0050]

[0051] where i is the type of flicker frequency, is the output frequency of the decoding correlation coefficient for the i-th type of flicker; is the frequency with the maximum correlation coefficient; max{i} is the total number of types of flicker frequencies; is the sum of the frequencies of the correlation coefficients for all types of flickers; ρ max is the proportion probability value of the obtained maximum correlation coefficient among all correlation coefficients.

[0052] Determine the gear values of the speed and angular velocity corresponding to the maximum correlation coefficient.

[0053] Furthermore, the process of continuous smoothing according to the determined target control command includes:

[0054] Through the linear combination of the basic speed and basic angular velocity determined by the maximum speed and maximum steering angle of the mobile wheeled robot and the additional speed and additional angular velocity determined by the proportion probability value of the maximum correlation coefficient among all correlation coefficients, the corresponding expected straight-line speed and angular velocity are obtained.

[0055] Furthermore, the fusion parameter is determined by fusing the balance factor determined by the state error generated by the user's control and the model error generated by the quality of the autonomous control model.

[0056] Specifically, the model error e is generated by the advantages and disadvantages of the autonomous control model MOD and the state error e generated by the imperfection of the subject's control EST , to determine the balance factor for fusion control. Among them, the model error e MOD is a constant value in the system, while the state error e EST is related to the distance between the robot controlled by the subject and the obstacle, and is updated in real time. The specific determination formula is as follows:

[0057]

[0058] where b and c are constant coefficients, determined according to the model error e MOD and the experimental effect, t is the sampling time, and d obs is the distance to the nearest obstacle.

[0059] A group of brain control speeds and the current speed of the robot, which are mapped from a period of EEG signals and consist of multiple v bci and ω bci signals, are iteratively fused multiple times to obtain a relatively accurate and stable fusion speed and fusion angular velocity at this moment, so as to realize the sharing control of the brain control drive control output and the robot's autonomous control:

[0060] x(k,t) = x(k - 1,t) + F(k,t)·(z(k,t) - x(k - 1,t));

[0061] where x(0,t) is the current state matrix of the robot [v robot , ω robot T , x(k,t) is the shared control output matrix [v sc , ω sc T , z(k,t) is the brain control output matrix [v bci , ω bci T , k is the iteration round at the sampling time, and F(k,t) is the fusion gain.

[0062] The fusion gain F is obtained by combining the balance factor P and the state error e EST :

[0063]

[0064] The balance factor P is used to balance the state error and model error of the system. At any sampling time, the initial value of the balance factor is:

[0065] P(k = 0,t) = e MOD ;​​​

[0066] At any moment, the balance factor P is updated after each round of iteration:

[0067] P(k + 1, t) = (1 - F(k, t))P(k, t).

[0068] Furthermore, the process of obtaining the local optimal solution includes:

[0069] Determine the speed sampling space according to the maximum speed constraint, sampling time constraint, and safety distance constraint of the mobile wheeled robot. Evaluate each set of discretized data in the speed sampling space through an evaluation function, and determine the set of data with the highest evaluation score as the local optimal solution; the evaluation function includes four evaluation items: target point, obstacle, moving speed, and shared speed. Specifically, the higher the evaluation function score when moving towards the target point, the higher the evaluation function score when moving away from the obstacle, the higher the evaluation function score when reaching the target point at a faster speed, and the higher the evaluation function score when approaching the shared speed and shared angular velocity.

[0070] Specifically, since the robot itself has maximum and minimum values for the magnitude of speed and the magnitude of steering angular velocity, the speed space that can be sampled must satisfy the speed limit. At this time, the speed space V s is:

[0071] V s = {(v, ω)|v ∈ [v max , v min , ω ∈ [ω max , ω min};

[0072] where v max and v min are the maximum and minimum speed limits of the robot respectively, and ω max and ω min are the maximum and minimum angular velocity limits of the robot respectively.

[0073] Since the output power of the robot itself is limited, when the robot is moving, restrictions are imposed on both linear acceleration and angular acceleration, that is, within each time window, the robot's ability to change the current state is limited. At this time, the speed space V that can be sampled d is:

[0074]

[0075] where v t and ω t are the linear velocity and angular velocity of the robot at the current sampling moment, and is the maximum linear acceleration and angular acceleration of the robot, and Δt is the time step.

[0076] When the robot encounters an obstacle during movement, it must stop before colliding with the obstacle and reduce its own speed to zero. This safety distance restricts the sampled velocity space V a as:

[0077]

[0078] where Dist(v, ω) is the closest distance between the current position of the robot and the obstacle.

[0079] Under the constraints of the above three constraints, the finally determined velocity sampling space under each time window is the intersection of the velocity spaces that can be sampled by the three. Let V t be the set of finally allowed sampling spaces, then V t is expressed as:

[0080] V t = V s ∩ V d ∩ V a ;

[0081] For the defined velocity sampling space V t , discretize its space. For each set of discretized (v, ω), simulate the trajectory of the next sampling time unit, and select the best trajectory and the corresponding space set (v, ω). For the selection criteria, combine the traditional DWA component and the relevant components of the shared fusion evaluation. The specific evaluation function should include the following indicators: the robot moves as close as possible to the target point; the robot moves as far away from the obstacle as possible; the robot reaches the target point as fast as possible; the moving speed of the robot is as close as possible to the shared speed and the shared angular velocity.

[0082] According to the above evaluation criteria, each evaluation function item is defined as:

[0083]

[0084] where d θ is the angle between the current orientation of the robot and the target point, d obs is the distance between the robot and the closest obstacle, d max is the obstacle distance constraint. If the distance between the robot and the obstacle during driving is greater than d max , it is regarded as a safe distance, and Dist(v, ω) is set to the safe distance d max , v sample and ω sample are the sampled velocity and the sampled angular velocity;

[0085] Combining each evaluation function term, the final evaluation function can be obtained:

[0086] G(v, ω)' = αHead(v, ω) + βDist(v, ω) + γVel(v, ω) + θSc(v, ω);

[0087] where α, β, γ, and θ are the proportion coefficients of each evaluation function.

[0088] As Figure 3 shown, in this embodiment, a continuous sharing control method for a brain-controlled wheeled robot system based on SSVEP is also provided, including:

[0089] Collect the electroencephalogram (EEG) signals generated by the user, decode the EEG signals to translate the user's intention, and obtain the target control instruction;

[0090] Perform continuous smoothing on the determined target control instruction to obtain the corresponding expected straight-line speed and angular velocity;

[0091] Based on the obtained corresponding expected straight-line speed and angular velocity, combined with the current state information of the mobile wheeled robot, and considering the surrounding environment state in real time to adjust the fusion parameters, obtain the shared speed and shared angular velocity after the fusion iteration is completed; where the current state information of the mobile wheeled robot includes the two-dimensional horizontal and vertical coordinates, forward speed, steering speed, and heading angle of the mobile wheeled robot;

[0092] Calculate the feasible speed window through the external environment information, combine the shared speed and shared angular velocity evaluation components, obtain the local optimal solution, and control the mobile wheeled robot according to the local optimal solution; where the external environment information includes the obstacle information and target point information of the environment.

[0093] Specifically, as Figure 2 shown, collect the electroencephalogram (EEG) signals generated by the user through a new type of stimulation panel. The new type of stimulation panel is a stimulation panel for collecting EEG signals using steady-state visual evoked potentials (SSVEPs). The stimulation array for activating the corresponding brain features in the stimulation panel is presented in the form of a two-dimensional stimulation surface, and the stimulation surface consists of 21 small strip rectangles in 3 rows and 7 columns; the horizontal dimension of the new type of stimulation panel represents the straight-line speed intention, including high-speed, medium-speed, and low-speed command signals, and the vertical dimension represents the steering angular velocity intention, including left and right directions with low, medium, and fast deflection speeds as well as no steering, a total of 7 command signals. The viewing angle between each small strip is approximately 4 degrees, and the stimulation frequency of the small strips diverges equidistantly from low to high in serial numbers from the inside to the outside, with a range of 6 - 10 Hz and an interval of 0.2 Hz. The visual stimulation phase difference between each adjacent small strip is π / 2.

[0094] Furthermore, decoding the EEG signals to translate the user's intention includes:

[0095] Decode the user's EEG signals to obtain the frequency with the maximum correlation coefficient;

[0096] Normalize the frequency with the maximum correlation coefficient to obtain the proportion probability value ρ of the maximum correlation coefficient among all correlation coefficients max ,

[0097]

[0098] where i is the type of flicker frequency, is the output frequency of the decoding correlation coefficient of the i-th flicker type; is the frequency with the maximum correlation coefficient; max{i} is the total number of types of flicker frequencies; is the sum of the frequencies of the correlation coefficients of all flicker types; ρ max is the proportion probability value of the obtained maximum correlation coefficient among all correlation coefficients.

[0099] Determine the gear values of the speed and angular velocity corresponding to the maximum correlation coefficient.

[0100] Furthermore, the continuous smoothing according to the determined target control instruction includes:

[0101] Obtain the corresponding expected straight-line speed and angular velocity through the linear combination of the maximum speed and maximum steering angle of the wheeled robot, the basic speed and basic angular velocity determined by the gear, and the additional speed and additional angular velocity determined by the proportion probability value of the maximum correlation coefficient among all correlation coefficients.

[0102] Specifically, based on the maximum speed and maximum steering angle of the robot, the linear combination of the basic speed and basic angular velocity determined by the gear and the additional speed and additional angular velocity determined by the maximum probability value ρ of the correlation coefficient max is used to determine the straight-line speed v corresponding to the final brain control output bci and the angular velocity ω bci :

[0103]

[0104] where v base and v addition represent the basic speed and the additional speed respectively, ω base and ω addition represent the basic angular velocity and the additional angular velocity respectively, and σ i , λ i , ο i , τ i are constant coefficients determined by the gear, and σ i , ο iDetermined directly by the command gear position, λ i , τ i Determined by the value range of each command gear position, v max is the maximum speed of the robot, ω max is the maximum steering angular velocity of the robot.

[0105] Furthermore, the fusion parameter is determined by fusing the balance factor determined by the state error generated by user control and the model error generated by the quality of the autonomous control model.

[0106] Specifically, the model error e MOD generated by the quality of the autonomous control model and the state error e EST generated by the imperfection of the subject's control are used to determine the balance factor for implementing fusion control. Among them, the model error e MOD is a constant value in the system, while the state error e EST is related to the distance between the robot controlled by the subject and the obstacle and is updated in real time. The specific determination formula is as follows:

[0107]

[0108] where b and c are constant coefficients, determined according to the model error e MOD and the experimental effect, t is the sampling time, and d obs is the distance to the nearest obstacle.

[0109] For a group of v bci and ω bci signals generated by mapping within a period of EEG signal, the brain-controlled speed and the current speed of the robot are iteratively fused multiple times to obtain a relatively accurate and stable fusion speed and fusion angular velocity at this moment to achieve shared control of the brain-controlled drive control output and the robot's autonomous control:

[0110] x(k,t) = x(k - 1,t) + F(k,t)·(z(k,t) - x(k - 1,t));

[0111] where x(0,t) is the robot's current state matrix [v robot , ω robot T , x(k,t) is the shared control output matrix [v sc , ω sc T , z(k,t) is the brain-controlled output matrix [v bci , ω bci T , k is the iteration round at the sampling time, and F(k,t) is the fusion gain.

[0112] ​​​The fusion gain F is obtained by combining the balance factor P and the state error e EST as follows:

[0113]

[0114] The balance factor P is used to balance the state error and the model error of the system. At any sampling moment, the initial value of the balance factor is:

[0115] P(k = 0,t) = e MOD ;

[0116] For any moment, the balance factor P is updated after each round of iteration:

[0117] P(k + 1,t) = (1 - F(k,t))P(k,t).

[0118] Furthermore, obtaining the local optimal solution includes:

[0119] Determine the speed sampling space according to the maximum speed constraint, sampling time constraint and safety distance constraint of the mobile wheeled robot. Evaluate each set of discretized data in the speed sampling space through an evaluation function, and the set of data with the highest evaluation score is determined as the local optimal solution; the evaluation function includes 4 evaluation items: target point, obstacle, moving speed and shared speed. Specifically, the higher the evaluation function score is when moving towards the target point, the higher the evaluation function score is when moving away from the obstacle, the higher the evaluation function score is when reaching the target point at a faster speed, and the higher the evaluation function score is when approaching the shared speed and shared angular velocity.

[0120] Specifically, since the robot itself has maximum and minimum values for the magnitude of speed and the magnitude of steering angular velocity, the speed space that can be sampled must satisfy the speed limit. At this time, the speed space V s is:

[0121] V s = {(v,ω)|v ∈ [v max ,v min ,ω ∈ [ω max ,ω min};

[0122] where, v max and v min are the maximum and minimum speed limits of the robot respectively, and ω max and ω min are the maximum and minimum angular velocity limits of the robot respectively.

[0123] Due to the limited output power of the robot itself, when the robot is moving, limitations are imposed on both linear acceleration and angular acceleration, that is, within each time window, the robot's ability to change its current state is limited. At this time, the velocity space V that can be sampled d is:

[0124]

[0125] where v t and ω t are the linear velocity and angular velocity of the robot at the current sampling moment, and are the maximum linear acceleration and angular acceleration of the robot, and Δt is the time step.

[0126] When the robot encounters an obstacle during movement, it must stop before colliding with the obstacle and reduce its own speed to zero. This safety distance restricts the sampled velocity space V a as:

[0127]

[0128] where Dist(v, ω) is the minimum distance between the current position of the robot and the obstacle.

[0129] Under the constraints of the above three constraints, the finally determined velocity sampling space per time window is the intersection of the velocity spaces that can be sampled by the three. Let V t be the set of finally allowed sampling spaces, then V t is expressed as:

[0130] V t = V s ∩V d ∩V a ;

[0131] For the defined velocity sampling space V t , discretize its space. For each set of discretized (v, ω), simulate the trajectory for the next sampling time unit, and select the best trajectory and the corresponding space set (v, ω). For the selection criteria, a combination of traditional DWA components and relevant components of shared fusion evaluation is used. The specific evaluation function should include the following indicators: the robot moves as much as possible towards the target point; the robot moves as far away from the obstacle as possible; the robot reaches the target point as fast as possible; the robot's moving speed is as close as possible to the shared speed and shared angular velocity.

[0132] According to the above evaluation criteria, each evaluation function item is defined as:

[0133]

[0134] where d θ is the angle between the current orientation of the robot and the target point, d obs is the distance between the robot and the nearest obstacle, d max is the obstacle distance constraint. If the distance between the robot and the obstacle is greater than d max during the driving process, it is regarded as a safe distance, and Dist(v, ω) is set to the safe distance d max , v sample and ω sample are the sampling speed and sampling angular velocity;

[0135] Combining each evaluation function term, the final evaluation function can be obtained:

[0136] G(v, ω)' = αHead(v, ω) + βDist(v, ω) + γVel(v, ω) + θSc(v, ω);

[0137] where α, β, γ, and θ are the proportion coefficients of each evaluation function.

[0138] Based on the brain-machine shared control, the present invention combines the brain-controlled output through continuous control and the autonomous control of the robot. On the one hand, the original discrete control input is changed into a continuous control input, realizing the continuous control of the brain-machine interface. On the other hand, shared control is realized. While ensuring that the user has a certain control right, it can assist in completing the driving task of the robot. Under the condition of ensuring safety, the brain-control control weight is maximized, which can not only ensure the participation of the user's autonomous consciousness but also enable the task to be completed as soon as possible. During the process of the user controlling the robot through the brain-machine interface, due to the instability of the control itself and the inaccuracy of decoding, there are certain errors in it. Coupled with the different advantages and disadvantages of the robot's autonomous control algorithm itself, when fusing the two, it is necessary to consider the weight proportion of each party. The present invention adjusts through the combination of the set model error and state error, and finds a stable and accurate control output in multiple iterations, increasing the control accuracy of the system. In terms of autonomous control, by combining the shared speed and angular velocity evaluation components, the brain-controlled output is added to the autonomous control regulation, improving the environmental adaptability and performance of the brain-controlled wheeled robot.

[0139] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any change or replacement that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A brain-controlled wheeled robot system based on SSVEP, characterized in that: include: Brain-computer interface module, smoothing module, shared control fusion module and dynamic window module; The brain-computer interface module is used to collect the EEG signals generated by the user, decode the EEG signals to translate the user's intentions, and obtain target control instructions; The smoothing module is used to perform continuous smoothing according to the determined target control instruction to obtain the corresponding expected straight-line speed and angular velocity; The shared control fusion module is used to adjust the fusion parameters in real time based on the obtained corresponding expected straight-line speed and angular velocity combined with the current state information of the mobile wheeled robot, while considering the surrounding environment state, to obtain the shared speed and shared angular velocity after the fusion iteration is completed; The dynamic window module is used to calculate a feasible speed window through external environment information, combine the shared speed and shared angular velocity evaluation components to obtain a local optimal solution, and control the mobile wheeled robot according to the local optimal solution.

2. The SSVEP-based brain-controlled wheeled robot system according to claim 1, characterized in that: The process of decoding the EEG signal and translating the user's intention includes: By decoding the user's EEG signal, the frequency with the maximum correlation coefficient is obtained; Normalizing the frequency of the maximum correlation coefficient to obtain a probability value of the maximum correlation coefficient corresponding to all correlation coefficients; Determine the gear value of speed and angular velocity corresponding to the maximum correlation coefficient.

3. The SSVEP-based brain-controlled wheeled robot system according to claim 1, characterized in that: The process of performing continuous smoothing according to the determined target control instruction includes: The corresponding expected straight-line speed and angular velocity are obtained by linearly combining the maximum speed and maximum steering angle of the mobile wheeled robot, the basic speed and basic angular velocity determined by the gear position, and the additional speed and additional angular velocity determined by the probability value of the maximum correlation coefficient corresponding to all correlation coefficients.

4. The SSVEP-based brain-controlled wheeled robot system according to claim 1, characterized in that: The fusion parameter is determined by fusing a balance factor determined by a state error generated by user control and a model error generated by the quality of the autonomous control model.

5. The SSVEP-based brain-controlled wheeled robot system according to claim 1, characterized in that: The process of obtaining the local optimal solution includes: The speed sampling space is determined according to the maximum speed constraint, sampling time constraint and safety distance constraint of the mobile wheeled robot. Each set of discretized data in the speed sampling space is evaluated by an evaluation function, and the set of data with the highest evaluation score is determined as the local optimal solution.

6. A continuous shared control method for a brain-controlled wheeled robot system based on SSVEP according to any one of claims 1 to 5, characterized in that: Collecting the EEG signals generated by the user, decoding the EEG signals to translate the user's intentions, and obtaining target control instructions; Continuously smoothing the target control command to obtain the corresponding expected straight-line speed and angular velocity; Based on the obtained corresponding expected straight speed and angular velocity combined with the current state information of the mobile wheeled robot, the fusion parameters are adjusted in real time considering the surrounding environment state to obtain the shared speed and shared angular velocity after the fusion iteration is completed; A feasible speed window is calculated through external environment information, and a local optimal solution is obtained by combining shared speed and shared angular velocity evaluation components. The mobile wheeled robot is controlled according to the local optimal solution.

7. The SSVEP-based brain-controlled wheeled robot system and its continuous shared control method as claimed in claim 6, characterized in that: Decoding the EEG signal to translate the user's intention includes: By decoding the user's EEG signal, the frequency with the maximum correlation coefficient is obtained; Normalizing the frequency of the maximum correlation coefficient to obtain a probability value of the maximum correlation coefficient corresponding to all correlation coefficients; Determine the gear value of speed and angular velocity corresponding to the maximum correlation coefficient.

8. The SSVEP-based brain-controlled wheeled robot system and its continuous shared control method as claimed in claim 6, characterized in that: Continuously smoothing according to the determined target control instruction includes: The corresponding expected straight-line speed and angular velocity are obtained by linearly combining the maximum speed and maximum steering angle of the mobile wheeled robot, the basic speed and basic angular velocity determined by the gear position, and the additional speed and additional angular velocity determined by the probability value of the maximum correlation coefficient corresponding to all correlation coefficients.

9. The SSVEP-based brain-controlled wheeled robot system and its continuous shared control method as claimed in claim 6, characterized in that: The fusion parameter is determined by fusing a balance factor determined by a state error generated by user control and a model error generated by the quality of the autonomous control model.

10. The SSVEP-based brain-controlled wheeled robot system and its continuous shared control method as claimed in claim 6, characterized in that: Obtaining the local optimal solution includes: The speed sampling space is determined according to the maximum speed constraint, sampling time constraint and safety distance constraint of the mobile wheeled robot. Each set of discretized data in the speed sampling space is evaluated by an evaluation function, and the set of data with the highest evaluation score is determined as the local optimal solution.

Images