Robot-assisted hand-eye coordination training system based on smooth eye tracking and guiding force field

Abstract

The present invention discloses a robot-assisted hand-eye coordination training system based on smooth eye tracking and guiding force field, comprising a virtual interactive scene module, a smooth eye tracking detection module, a robot-assisted interception module and an impact force rendering module. The virtual interactive scene module can generate a virtual interactive scene with a virtual moving object and a virtual handle agent; the smooth eye tracking detection module collects eye movement signals when the user tracks the virtual moving object to detect smooth eye tracking events. The robot-assisted interception module estimates the direction of movement of the virtual moving object and generates an interception guiding force field, thereby generating an auxiliary force to assist the user in interception. The impact force rendering module collects the kinematic information of the robot handle for collision detection and calculates the impact force according to the impact force calculation model after the collision is detected, and controls the motor to generate force feedback to the user's hand. This improves the user's enthusiasm and interest in participating in hand-eye coordination training and enhances the immersion of the system.

Description

Technical Field

[0001] The present invention relates to a robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field, belonging to the fields of rehabilitation training and motor learning. Background Art

[0002] Stroke is an acute vascular disease caused by the sudden rupture of a cerebral blood vessel, temporarily or permanently blocking it, preventing blood flow to the brain and causing damage to brain tissue. It is characterized by high morbidity, disability, mortality, recurrence, and economic burden, and has long been considered one of the leading causes of death and disability worldwide. Depending on the location of the lesion, stroke patients may experience limb dysfunction, with hand-eye coordination impairment being a common limb dysfunction. Hand-eye coordination is the ability of the brain to respond promptly to information received by the eyes and transmitted via nerves, sending control signals to control hand muscles to perform corresponding movements. It reflects the coordination and stability of the human nervous system, as well as the coordination between the hand and the eye. Although tissue damage caused by stroke is irreversible, relevant research shows that the brain's ability to remodel can restore some limb function. Targeted rehabilitation training, after conventional treatment, can play a vital role in the patient's recovery. Traditional rehabilitation training methods mainly rely on professional rehabilitation therapists to lead patients in rehabilitation training. During this training process, patients can only passively follow the therapist's arrangements for rehabilitation training. The training process is boring and tedious, and the training results and plans rely more on the therapist's subjective judgment. This requires a high level of professionalism from rehabilitation therapists. Faced with a large patient population, professional therapists are often in short supply. In order to alleviate the extremely asymmetric situation between the number of therapists and rehabilitation patients, people have turned their attention to the research and development of robot-assisted upper limb training systems to replace therapists' accompanying training of patients. Upper limb training robot systems generally consist of the robot itself and a supporting virtual interactive scene. The robot provides power to the patient, and the virtual interactive scene provides different training tasks and visual feedback.

[0003] Research has found that when robots assist users with upper limb training, incorporating the user's motor intention into the training can effectively stimulate the brain's motor-related cortex, accelerate neural remodeling, and improve training efficiency. As a crucial sensory organ, the eyes can capture a wealth of external information, and eye movements can reflect a person's motor intentions. They not only receive information but also serve as an input channel for information reflecting their motor intentions. Before manipulating a moving object, the eyes will first track it through smooth tracking, acquiring relevant motion information and performing the corresponding manipulation. Compared to acquiring the user's motor intention through EEG or EMG, using eye movements is more convenient and simple, requiring only an eye tracker. EEG requires applying EEG cream and wearing an EEG cap, while EMG requires electrodes to be attached to the target muscles. Summary of the Invention

[0004] To solve the above problems, the present invention discloses a robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field. By collecting the user's eye movement signals, smooth eye tracking events are detected therefrom, thereby estimating the movement direction of virtual moving objects in the virtual interactive scene to obtain the user's movement intention, and generating an interception guiding force field to assist the user in completing the training task, thereby realizing the training of hand-eye coordination ability.

[0005] In order to achieve the above technical objectives, the present invention will adopt the following technical solutions:

[0006] A robot-assisted hand-eye coordination training system based on smooth eye tracking and a guiding force field is used to assist users in using the robot handle of an upper limb rehabilitation robot for rehabilitation training. The system includes a virtual interactive scene module, a smooth eye tracking detection module, a robot-assisted interception module, and an impact force rendering module.

[0007] The virtual interaction scene module can generate a virtual interaction scene for hand-eye coordination training; the virtual interaction scene has a virtual moving object that can provide visual motion stimulation to the user and a handle virtual agent that matches the movement of the robot handle;

[0008] The smooth eye tracking detection module calculates the user's eye angular velocity by collecting eye movement signals when the user tracks a virtual moving object in a virtual interactive scene, and classifies eye movement events according to the calculated eye angular velocity, thereby detecting smooth eye tracking events, and transmitting the detected smooth eye tracking events to the robot-assisted interception module;

[0009] The robot-assisted interception module estimates the movement direction of the virtual moving object in the virtual interactive scene through the eye movement smooth tracking events detected by the eye movement smooth tracking detection module, and generates an interception guidance force field to generate auxiliary force to assist the user in pushing the robot handle;

[0010] The impact force rendering module obtains the position of the handle virtual agent in the virtual interaction scene by collecting the kinematic information of the robot handle, and then determines whether the handle virtual agent collides with the virtual moving object by comparing the position of the virtual moving object in the virtual interaction scene and the position of the handle virtual agent; when the judgment result shows that the handle virtual agent successfully intercepts the virtual moving object, the impact force generated when the handle virtual agent collides with the virtual moving object is calculated to apply a feedback force matching the impact force to the user's hand.

[0011] As a further improvement of the above technical solution, the eye movement smooth tracking detection module includes an eye movement signal acquisition module, an eye movement signal preprocessing module, an eye movement angular velocity calculation module, and an eye movement event classification module, wherein:

[0012] The eye movement signal acquisition module is used to collect the user's eye movement signals in real time and transmit them to the eye movement signal preprocessing module;

[0013] The eye movement signal preprocessing is used to remove invalid signals from the eye movement signal transmitted by the eye movement signal acquisition module and perform filtering and denoising processing;

[0014] The eye movement angular velocity calculation module can calculate the eye movement angular velocity according to the eye movement signal; the calculation formula of the eye movement angular velocity ω is as follows:

[0015]

[0016] Among them, (x i ,y i ) is the coordinate of the current sampling point, (x i-1 ,y i-1 ) is the coordinate of the previous sampling point, a is the width of the display interface, b is the height of the display interface, l is the vertical distance from the user's eyes to the display interface; t is the sampling period of the eye movement signal;

[0017] The eye movement event classification module is built based on the IVVT classification method. It can classify user eye movement events according to eye movement angular velocity to detect eye movement smooth tracking events in user eye movement events.

[0018] As a further improvement of the above technical solution, the eye movement signal acquisition module uses an eye tracker to collect the user's eye movement signals.

[0019] As a further improvement of the above technical solution, the eye movement event classification module is preset with two speed thresholds ω th_fix 、ω th_sac ; When the eye angular velocity calculation module calculates the eye angular velocity ω of the current sampling point to meet ω th_fix <ω<ω th_sac , the current sampling point is marked as smooth tracking.

[0020] As a further improvement of the above technical solution, the robot-assisted interception module includes a motion direction estimation module and an interception guidance force field generation module, wherein:

[0021] The motion direction estimation module first estimates the motion trajectory (x, y) of the virtual moving object in the virtual interactive scene using a univariate linear regression method, then obtains the motion direction of the virtual moving object based on the estimated motion trajectory, and transmits the obtained motion direction of the virtual moving object to the interception guidance force field generation module; the motion trajectory (x, y) of the virtual moving object satisfies:

[0022] y=α+βx

[0023] Among them, α and β are the regression constant and regression coefficient, respectively, which are obtained by least square fitting, x is the horizontal coordinate of the virtual moving object, and y is the vertical coordinate of the virtual moving object;

[0024] The interception guidance force field generation module generates an interception guidance force field based on the estimated motion direction of the virtual moving object, generating an auxiliary force to assist the user in operating the robot handle to intercept the virtual moving object in the virtual interactive scene; the interception guidance force field is expressed as follows:

[0025]

[0026] Among them, F assist To intercept the auxiliary force generated by the guiding force field, F m is the maximum auxiliary force generated by the interception guidance force field, d is the vertical distance from the handle virtual agent to the motion trajectory of the virtual moving object, k is the auxiliary force coefficient, d0 is the vertical distance between the boundary between the interception guidance force field and the free interception area relative to the motion trajectory of the virtual moving object, and d m The critical distance for achieving maximum assist force.

[0027] As a further improvement of the above technical solution, the impact force rendering module includes a robot handle kinematic information acquisition module, a collision detection module, an impact force calculation module and an execution module, wherein:

[0028] The robot handle kinematic information acquisition module collects the handle kinematic information when the user's upper limbs operate the robot handle through internal sensors;

[0029] The collision detection module uses the collected controller kinematic information to determine whether the controller has successfully intercepted the virtual moving object in the virtual interactive scene; if the detection result shows that the controller has collided with the virtual moving object, it means that the controller has successfully intercepted the virtual moving object;

[0030] The impact force calculation module calculates the impact force generated when the handle collides with the virtual moving object using an impact force calculation model; the impact force calculation model is expressed as follows:

[0031]

[0032] Where F is the impact force, m b is the mass of the virtual moving object in the virtual interactive scene, m p is the agent quality of the handle virtual agent in the virtual interaction scene, v b0 is the speed of the virtual moving object in the virtual interactive scene before the collision, v p0 is the agent velocity of the handle virtual agent in the virtual interaction scene before the collision, and Δt is the collision duration;

[0033] The motor execution module generates a motor execution control signal according to the impact force calculated by the impact force calculation module, and controls the operation of the motor to generate a feedback force that reacts to the user's hand and matches the aforementioned impact force.

[0034] As a further improvement of the above technical solution, the virtual interaction scene module includes a training scene generation module and a feedback module, wherein:

[0035] The training scene generation module generates a virtual interactive scene for hand-eye coordination training based on the Pygame platform;

[0036] The feedback module is used to provide the user with a virtual moving object with visual motion stimulation in a virtual interactive environment, and to display a handle virtual agent that matches the movement of the robot handle.

[0037] As a further improvement to the above technical solution, the robot-assisted hand-eye coordination training system is implemented according to the following steps:

[0038] Step 1: collecting eye movement signals when the user tracks a virtual moving object in a virtual interactive scene and performing preprocessing, eye movement angular velocity calculation, and eye movement event classification to detect smooth eye movement tracking events;

[0039] Step 2: Estimate the direction of movement of the virtual moving object through eye movement smooth tracking events and generate an interception guidance force field to assist the user in pushing the robot handle to intercept the virtual moving object in the virtual interactive scene to complete the training task. At the same time, collect the kinematic information of the robot handle in real time for collision detection to detect whether the interception is successful. When the virtual moving object is successfully intercepted, the impact is calculated through the impact force calculation model and force feedback is generated to the user's hand through motor control.

[0040] As a further improvement to the above technical solution, in step 1, the eye movement angular velocity calculation formula is:

[0041]

[0042] Among them, (x i ,y i ) is the coordinate of the current sampling point, (x i-1 ,y i-1 ) is the coordinate of the previous sampling point, a is the width of the display interface, b is the height of the display interface, l is the vertical distance from the user's eyes to the display interface; t is the sampling period of the eye movement signal;

[0043] The IVVT classification method is used to classify eye movement events. The specific method is as follows:

[0044] Set two speed thresholds ω th_fix and ω th_sac , when the eye angular velocity is greater than the velocity threshold ω th_fix At the same time, it is less than the speed threshold ω th_sac , the current sampling point is marked as smooth tracking.

[0045] As a further improvement to the above technical solution, in step 2, the interception guidance force field is expressed as follows:

[0046]

[0047] Among them, F assist To intercept the auxiliary force generated by the guiding force field, F m is the maximum auxiliary force generated by the interception guidance force field, d is the vertical distance from the handle virtual agent to the motion trajectory of the virtual moving object, k is the auxiliary force coefficient, d0 is the vertical distance between the boundary between the interception guidance force field and the free interception area relative to the motion trajectory of the virtual moving object, and d m The critical distance to achieve maximum assist force;

[0048] The impact force calculation model is expressed as follows:

[0049]

[0050] Where F is the impact force, m bis the mass of the virtual moving object in the virtual interactive scene, m p is the agent quality of the handle virtual agent in the virtual interaction scene, v b0 is the speed of the virtual moving object in the virtual interactive scene before the collision, v p0 is the agent velocity of the handle virtual agent in the virtual interaction scene before the collision, and Δt is the collision duration.

[0051] Based on the above technical objectives, the present invention has the following advantages over the prior art:

[0052] (1) By detecting the user's eye movement smooth tracking events, the system can easily estimate the movement direction of virtual moving objects in the virtual interactive scene and thus obtain the user's movement intention.

[0053] (2) The system generates an interception guidance force field to produce auxiliary force to help users with poor hand-eye coordination to better and more accurately intercept virtual moving objects, thereby exercising their hand-eye coordination ability.

[0054] (3) The system uses impact force rendering to generate impact force feedback when the user successfully intercepts a virtual moving object, which increases the realism of the hand-eye coordination training system and the immersion of the training process.

[0055] (4) The hand-eye coordination training system uses virtual interactive scenes to increase users' enthusiasm and interest in participating in hand-eye coordination training. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a system framework diagram of the present invention;

[0057] Figure 2 Schematic diagram of the interception guidance force field. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. The following description of at least one exemplary embodiment is actually only illustrative and in no way limits the present invention and its application or use. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of the present invention. Unless otherwise specified, the relative arrangement of components and steps, expressions and numerical values ​​described in these embodiments do not limit the scope of the present invention. Technologies, methods and equipment known to ordinary technicians in the relevant fields may not be discussed in detail, but where appropriate, the technologies, methods and equipment should be considered as part of the authorization specification. In all examples shown and discussed here, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of the exemplary embodiments may have different values.

[0059] The training tasks in this invention typically refer to complex sports tasks in daily life that require hand-eye coordination, such as playing table tennis or badminton. Here, the therapist / technician provides the target movement, and the subject completes the training task by tracking the target's movement with their eyes and pushing the robot's handle to intercept the target.

[0060] like Figure 1 As shown, the robot-assisted hand-eye coordination training system based on eye movement smooth tracking and force feedback provided by the embodiment of the present invention includes the following steps:

[0061] (1) Collect the user's eye movement signals when tracking virtual moving objects in the virtual interactive scene and detect smooth eye movement tracking events from them.

[0062] The user's eye movement signals collected by the eye tracker are subjected to invalid signal elimination and Kalman filtering, and then the eye movement angular velocity is calculated. Then, eye movement events are classified according to the eye movement angular velocity to detect eye movement smooth tracking events.

[0063] The eye tracker used was the Pupil Core eye tracker from Pupil Labs in Berlin, Germany.

[0064] The eye movement angular velocity is calculated as follows:

[0065]

[0066] Among them, θ is the rotation angle of the current sampling point of the user's eye movement signal relative to the previous sampling point, (x i ,y i ) is the coordinate of the current sampling point, (x i-1 ,yi-1 ) is the coordinate of the previous sampling point, a is the width of the display interface, b is the height of the display interface, and l is the vertical distance from the user's eyes to the display interface;

[0067] The angular velocity of the user's eye movement signal is calculated by the rotation angle of the current sampling point relative to the previous sampling point, that is, the eye movement angular velocity is obtained:

[0068]

[0069] Where ω is the eye movement angular velocity, θ is the rotation angle of the collected user's eye movement signal at the current sampling point relative to the previous sampling point, and t is the sampling period of the eye movement signal.

[0070] Eye movement events were classified using the IVVT classification method, as follows:

[0071] Set two speed thresholds ω th_fix and ω th_sac , when the eye angular velocity is greater than the velocity threshold ω th_fix At the same time, it is less than the speed threshold ω th_sac , the current sampling point is marked as smooth tracking.

[0072] (2) Estimate the direction of motion of the virtual moving object and generate an interception guidance force field, thereby generating an auxiliary force to assist the user in intercepting; Figure 2 Directs the force field for the resulting interceptor.

[0073] The movement direction of the virtual moving object is estimated based on the detected eye movement smooth tracking events to obtain the user's movement intention, thereby generating an interception guidance force field and generating an auxiliary force to assist the user in pushing the robot handle, so that the handle virtual agent in the virtual interaction scene can intercept the virtual moving object.

[0074] Motion direction estimation: The motion direction of the virtual moving object in the virtual interactive scene is estimated by detecting the smooth eye tracking event. The estimation method is as follows:

[0075] Use univariate linear regression to estimate the motion trajectory of virtual moving objects in virtual interactive scenes:

[0076] y=α+βx

[0077] Among them, α and β are the regression constant and regression coefficient respectively, which are obtained by least square fitting (the fitting data comes from the coordinates of the above-mentioned sampling points), x is the horizontal coordinate of the virtual moving object, and y is the vertical coordinate of the virtual moving object.

[0078] The moving direction of the virtual moving object can be obtained according to the estimated moving trajectory.

[0079] Intercept guidance force field generation: Based on the estimated direction of motion of the object, an intercept guidance force field is generated to generate auxiliary force to assist the user in operating the upper limb rehabilitation robot (ArmMotus of Shanghai Fourier Intelligent Technology Co., Ltd.) TM M2) of the robot handle enables the handle virtual agent in the virtual interactive scene to intercept the virtual moving object; the interception guidance force field is expressed as follows:

[0080]

[0081] Among them, F assist To intercept the auxiliary force generated by the guiding force field, F m is the maximum auxiliary force generated by the interception guidance force field, d is the vertical distance from the handle virtual agent to the motion trajectory of the virtual moving object, k is the auxiliary force coefficient, d0 is the critical distance with or without auxiliary force (that is, the vertical distance between the boundary between the interception guidance force field and the free interception area relative to the motion trajectory of the virtual moving object), d m To achieve the maximum assist force F m The critical distance when the maximum assist force F m The vertical distance from the handle virtual agent to the motion trajectory of the virtual moving object).

[0082] (3) Collect the kinematic information of the robot handle for collision detection and calculate the impact force according to the impact force calculation model after detecting the collision (i.e. successfully intercepting the virtual moving object) and control the motor to generate force feedback to the user's hand.

[0083] When the user pushes the robot handle, prompting the handle virtual agent in the virtual interactive scene to intercept the virtual moving object, the upper limb rehabilitation robot (ArmMotus of Shanghai Fourier Intelligent Technology Co., Ltd.) TM Internal sensors in the robot's M2 collect real-time kinematic information about the user's robot handle for collision detection. When the handle's virtual agent successfully intercepts a virtual moving object (i.e., a collision is detected) in the virtual interaction scene, the impact force is calculated using the collision calculation model. Then, using the DynaLinkHS.CmdJointKineticControl control method in the SDK (FFTAICommunicationLib) library in the upper limb rehabilitation robot, the motor is controlled to generate the impact force and feedback it to the user's hand.

[0084] Specifically, the impact force calculation model is expressed as follows:

[0085]

[0086] Where F is the impact force, m b is the mass of the virtual moving object in the virtual interactive scene, m pis the agent quality of the handle virtual agent in the virtual interaction scene, v b0 is the speed of the virtual moving object in the virtual interactive scene before the collision, v p0 is the agent velocity of the handle virtual agent in the virtual interaction scene before the collision, and Δt is the collision duration.

[0087] (4) Initialize the training task scenario

[0088] A two-dimensional table tennis virtual interactive scene is used as the task training scene, and the positions and velocities of the table tennis ball and racket are initialized. In this case, the table tennis ball in the virtual interactive scene is the aforementioned virtual moving object, while the racket is the aforementioned virtual proxy for the handle. Other training scenes can also be used. Of course, the present invention may be more suitable for ball game training scenarios. Besides the aforementioned table tennis virtual interactive scene, tennis, badminton, and other sports can also be used.

[0089] (5) Conduct hand-eye coordination training

[0090] Through extensive and long training sessions, the subjects' hand-eye coordination abilities are continuously strengthened. During training, the speed and direction of the table tennis ball are randomly changed to prevent the subjects from adapting and thus reducing the effectiveness of the training.

Claims

1. A robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field, used to assist users in using the robot handle of an upper limb rehabilitation robot for rehabilitation training, characterized in that: It includes a virtual interactive scene module, an eye movement smooth tracking detection module, a robot-assisted interception module, and an impact force rendering module, among which: The virtual interaction scene module can generate a virtual interaction scene for hand-eye coordination training; the virtual interaction scene has a virtual moving object that can provide visual motion stimulation to the user and a handle virtual agent that matches the movement of the robot handle; The smooth eye tracking detection module calculates the user's eye angular velocity by collecting eye movement signals when the user tracks a virtual moving object in a virtual interactive scene, and classifies eye movement events according to the calculated eye angular velocity, thereby detecting smooth eye tracking events, and transmitting the detected smooth eye tracking events to the robot-assisted interception module; The robot-assisted interception module estimates the movement direction of the virtual moving object in the virtual interactive scene through the eye movement smooth tracking events detected by the eye movement smooth tracking detection module, and generates an interception guidance force field to generate auxiliary force to assist the user in pushing the robot handle; The impact force rendering module obtains the position of the handle virtual agent in the virtual interaction scene by collecting the kinematic information of the robot handle, and then determines whether the handle virtual agent collides with the virtual moving object by comparing the position of the virtual moving object in the virtual interaction scene and the position of the handle virtual agent; when the judgment result shows that the handle virtual agent successfully intercepts the virtual moving object, the impact force generated when the handle virtual agent collides with the virtual moving object is calculated to apply a feedback force matching the impact force to the user's hand.

2. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 1 is characterized in that: The eye movement smooth tracking detection module includes an eye movement signal acquisition module, an eye movement signal preprocessing module, an eye movement angular velocity calculation module, and an eye movement event classification module, wherein: The eye movement signal acquisition module is used to collect the user's eye movement signals in real time and transmit them to the eye movement signal preprocessing module; The eye movement signal preprocessing is used to remove invalid signals from the eye movement signal transmitted by the eye movement signal acquisition module and perform filtering and denoising processing; The eye movement angular velocity calculation module can calculate the eye movement angular velocity according to the eye movement signal; the calculation formula of the eye movement angular velocity ω is as follows: Among them, (x i ,y i ) is the coordinate of the current sampling point, (x i-1 ,y i-1 ) is the coordinate of the previous sampling point, a is the width of the display interface, b is the height of the display interface, l is the vertical distance from the user's eyes to the display interface; t is the sampling period of the eye movement signal; The eye movement event classification module is built based on the IVVT classification method. It can classify user eye movement events according to eye movement angular velocity to detect eye movement smooth tracking events in user eye movement events.

3. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 2 is characterized in that: The eye movement signal acquisition module uses an eye tracker to acquire the user's eye movement signals.

4. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 3 is characterized in that: The eye movement event classification module is preset with two speed thresholds ω th_fix 、ω th_sac ; When the eye angular velocity calculation module calculates the eye angular velocity ω of the current sampling point to meet ω th_fix <ω<ω th_sac , the current sampling point is marked as smooth tracking.

5. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 2 is characterized in that: The robot-assisted interception module includes a motion direction estimation module and an interception guidance force field generation module, wherein: The motion direction estimation module first estimates the motion trajectory (x, y) of the virtual moving object in the virtual interactive scene using a univariate linear regression method, then obtains the motion direction of the virtual moving object based on the estimated motion trajectory, and transmits the obtained motion direction of the virtual moving object to the interception guidance force field generation module; the motion trajectory (x, y) of the virtual moving object satisfies: y=α+βx Among them, α and β are the regression constant and regression coefficient, respectively, which are obtained by least square fitting, x is the horizontal coordinate of the virtual moving object, and y is the vertical coordinate of the virtual moving object; The interception guidance force field generation module generates an interception guidance force field based on the estimated motion direction of the virtual moving object, generating an auxiliary force to assist the user in operating the robot handle to intercept the virtual moving object in the virtual interactive scene; the interception guidance force field is expressed as follows: Among them, F assist To intercept the auxiliary force generated by the guiding force field, F m is the maximum auxiliary force generated by the interception guidance force field, d is the vertical distance from the handle virtual agent to the motion trajectory of the virtual moving object, k is the auxiliary force coefficient, d0 is the vertical distance between the boundary between the interception guidance force field and the free interception area relative to the motion trajectory of the virtual moving object, and d m The critical distance for achieving maximum assist force.

6. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 5 is characterized in that: The impact force rendering module includes a robot handle kinematic information acquisition module, a collision detection module, an impact force calculation module, and an execution module, wherein: The robot handle kinematic information acquisition module collects the handle kinematic information when the user's upper limbs operate the robot handle through internal sensors; The collision detection module uses the collected controller kinematic information to determine whether the controller has successfully intercepted the virtual moving object in the virtual interactive scene; if the detection result shows that the controller has collided with the virtual moving object, it means that the controller has successfully intercepted the virtual moving object; The impact force calculation module calculates the impact force generated when the handle collides with the virtual moving object using an impact force calculation model; the impact force calculation model is expressed as follows: Where F is the impact force, m b is the mass of the virtual moving object in the virtual interactive scene, m p is the agent quality of the handle virtual agent in the virtual interaction scene, v b0 is the speed of the virtual moving object in the virtual interactive scene before the collision, v p0 is the agent velocity of the handle virtual agent in the virtual interaction scene before the collision, and Δt is the collision duration; The motor execution module generates a motor execution control signal according to the impact force calculated by the impact force calculation module, and controls the operation of the motor to generate a feedback force that reacts to the user's hand and matches the aforementioned impact force.

7. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 4 is characterized in that: The virtual interaction scene module includes a training scene generation module and a feedback module, wherein: The training scene generation module generates a virtual interactive scene for hand-eye coordination training based on the Pygame platform; The feedback module is used to provide the user with a virtual moving object with visual motion stimulation in a virtual interactive environment, and to display a handle virtual agent that matches the movement of the robot handle.

8. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 1 is characterized in that: The robot-assisted hand-eye coordination training system is performed according to the following steps: Step 1: collecting eye movement signals when the user tracks a virtual moving object in a virtual interactive scene and performing preprocessing, eye movement angular velocity calculation, and eye movement event classification to detect smooth eye movement tracking events; Step 2: Estimate the direction of movement of the virtual moving object through eye movement smooth tracking events and generate an interception guidance force field to assist the user in pushing the robot handle to intercept the virtual moving object in the virtual interactive scene to complete the training task. At the same time, collect the kinematic information of the robot handle in real time for collision detection to detect whether the interception is successful. When the virtual moving object is successfully intercepted, the impact is calculated through the impact force calculation model and force feedback is generated to the user's hand through motor control.

9. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 8, characterized in that: In step 1, the formula for calculating eye angular velocity is: Among them, (x i ,y i ) is the coordinate of the current sampling point, (x i-1 ,y i-1 ) is the coordinate of the previous sampling point, a is the width of the display interface, b is the height of the display interface, l is the vertical distance from the user's eyes to the display interface; t is the sampling period of the eye movement signal; The IVVT classification method is used to classify eye movement events. The specific method is as follows: Set two speed thresholds ω th_fix and ω th_sac , when the eye angular velocity is greater than the velocity threshold ω th_fix At the same time, it is less than the speed threshold ω th_sac , the current sampling point is marked as smooth tracking.

10. The robot-assisted hand-eye coordination training system based on eye movement smooth tracking and guiding force field according to claim 8, characterized in that: In step 2, the interception guidance force field is expressed as follows: Among them, F assist To intercept the auxiliary force generated by the guiding force field, F m is the maximum auxiliary force generated by the interception guidance force field, d is the vertical distance from the handle virtual agent to the motion trajectory of the virtual moving object, k is the auxiliary force coefficient, d0 is the vertical distance between the boundary between the interception guidance force field and the free interception area relative to the motion trajectory of the virtual moving object, and d m The critical distance to achieve maximum assist force; The impact force calculation model is expressed as follows: Where F is the impact force, m b is the mass of the virtual moving object in the virtual interactive scene, m p is the agent quality of the handle virtual agent in the virtual interaction scene, v b0 is the speed of the virtual moving object in the virtual interactive scene before the collision, v p0 is the agent velocity of the handle virtual agent in the virtual interaction scene before the collision, and Δt is the collision duration.

Images