Control method, device and equipment of bronchoscope auxiliary robot

Abstract

The invention discloses a control method, device and equipment for a bronchoscope auxiliary robot, and relates to the technical field of auxiliary robot control. The method includes the steps that a mechanical arm tail end path point is obtained and converted into a joint angle sequence through inverse kinematics; a B-spline curve is adopted to plan a trajectory, an optimization model with the total time and impact minimized is established, an optimal time sequence is solved through an improved sparrow algorithm, an optimized trajectory is obtained, and the mechanical arm is controlled to move; endoscope images and virtual navigation images are collected, structural similarity measurement is established based on regional gray values and rigid transformation after preprocessing, and rolling angle deviation is solved by using an improved particle swarm algorithm and registration is carried out. According to the invention, the impact of the mechanical arm can be reduced, the registration precision is improved, and the rolling angle deviation between the navigation image and the endoscope image is eliminated.

Description

Technical Field

[0001] This invention relates to the field of assistive robot control technology, and more specifically to a control method, apparatus, and equipment for a bronchoscopy assistive robot. Background Technology

[0002] Bronchoscopic biopsy is an important method for diagnosing lung diseases such as lung cancer. Traditional bronchoscopy has problems such as complex operation, steep learning curve, and reliance on the doctor's experience for positioning accuracy. To address these issues, electromagnetic navigation bronchoscopy (ENB) technology has been introduced, which assists doctors in navigating to the lesion by tracking the position of the bronchoscope tip in real time.

[0003] However, existing electromagnetic navigation bronchoscopic surgical systems and methods still have the following technical problems: Decreased registration accuracy due to robotic arm movement: During preoperative spatial registration and the operation itself, the starting, stopping, or movement of the robotic arm joints generates minute vibrations and impacts. These disturbances directly affect the registration accuracy of the electromagnetic navigation system, leading to an increase in theoretical registration error, and consequently impacting the accuracy and safety of surgical navigation.

[0004] Roll angle deviation between navigation and endoscopic images: After initial registration, due to the use of a five-degree-of-freedom electromagnetic sensor and other factors, there is a deviation between the virtual navigation image (based on preoperative CT reconstruction) and the real-time endoscopic image in the roll angle dimension (i.e., the rotation angle around the catheter axis). This deviation makes it difficult for surgeons to quickly and accurately align the catheter tip with the target bronchial opening when encountering bronchial bifurcation, reducing surgical efficiency and increasing operational difficulty.

[0005] Therefore, there is an urgent need for a control method for bronchoscopy-assisted robots that can overcome the above-mentioned defects. Summary of the Invention

[0006] The purpose of this invention is to provide a control method, device, and equipment for a bronchoscopy-assisted robot. First, trajectory planning is performed in joint space using B-spline curves, and an optimization mathematical model is established with the goal of minimizing total motion time and joint impact. An improved sparrow algorithm is used to solve for the optimal time series, and the initial trajectory is time-redistributed to obtain the optimized joint motion trajectory. This effectively suppresses vibration and impact during the start-up, shutdown, and movement of the robotic arm, significantly improving the spatial registration accuracy of the electromagnetic navigation system. Based on this, virtual navigation images and real-time endoscope images are further acquired. A similarity metric is established based on regional grayscale values ​​and a rigidity transformation model. An improved particle swarm optimization algorithm is used to globally optimize and solve for the roll angle deviation, and then registration is performed to ensure that the roll angle directions of the two images are consistent, providing an accurate attitude reference for subsequent precise catheter guidance. This method achieves high-precision, low-vibration control of the bronchoscopy-assisted robot through the synergy of trajectory optimization and image registration.

[0007] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a control method for a bronchoscopy-assisted robot, the method comprising: Obtain multiple Cartesian space path points that the end effector of the bronchoscope-assisted robot needs to pass through; Inverse kinematics is used to convert each path point into a sequence of joint angles for each joint of the robotic arm. Based on the joint angle sequence, B-spline curves are used to plan the trajectory of each joint in the joint space to generate the initial joint motion trajectory. An optimization mathematical model was established with the goal of minimizing total motion time and joint impact. An improved sparrow algorithm was used to solve the optimization mathematical model and obtain the optimal time series. The initial joint motion trajectory is redistributed over time using the optimal time series to obtain the optimized joint motion trajectory; Control the robotic arm to move according to the optimized joint motion trajectory; The bronchoscope on the bronchoscope-assisted robot acquires endoscopic images in real time and obtains the corresponding virtual navigation images generated by the electromagnetic navigation system. Preprocess the virtual navigation image and endoscope image; Based on the regional grayscale values ​​of the preprocessed image, a rigid transformation model is used to establish a similarity measure for image registration. The rigid transformation model includes translation parameters, rotation parameters, and scaling parameters, and the structural similarity index is used as the similarity measure. An improved particle swarm optimization algorithm was used to globally optimize the similarity metric, and the roll angle deviation between the virtual navigation image and the endoscope image was solved. Based on the calculated roll angle deviation, the virtual navigation image and the endoscope image are registered to ensure that their roll angle directions are consistent.

[0008] In some embodiments, improvements to the Sparrow Algorithm include: An elite reverse learning strategy was used to generate an initial sparrow population. During the iteration process, the population fitness variance is calculated; When the fitness variance is lower than a preset threshold, Cauchy mutation is performed on the current best sparrow individual; When the fitness variance exceeds a preset threshold, Gaussian mutation is performed on elite sparrow individuals.

[0009] In some embodiments, an optimization mathematical model is established with the objective of minimizing total motion time and joint impact, including: Use the travel time between adjacent path points as a design variable; The limits of angular velocity, angular acceleration, and angular jerk of each joint of the robotic arm are used as constraints. The fitness function is the weighted sum of the total motion time of all path segments and the impact cost function.

[0010] In some embodiments, preprocessing of the virtual navigation image and the endoscope image includes: The virtual navigation image and endoscope image are converted to grayscale images, and the grayscale virtual navigation image is then inverted.

[0011] In some embodiments, an improved particle swarm optimization algorithm is used to globally optimize the similarity metric to solve for the roll angle deviation between the virtual navigation image and the endoscopic image, including: Initialize the particle swarm, where the position vector of each particle includes the image scaling ratio, rotation angle, horizontal offset, and vertical offset; Calculate the fitness value for each particle, which is based on a similarity metric; Update the individual historical best position of each particle and the global historical best position of the particle swarm; During the iteration process, the learning factor is dynamically adjusted so that the individual learning factor decreases as the number of iterations increases, while the social learning factor increases as the number of iterations increases. A differential mutation strategy is introduced to generate mutation vectors, and the target vector and mutation vectors are cross-operated to generate experimental vectors; The particle position is updated based on the experimental vector, and the iteration is repeated until the termination condition is met. The globally optimal position is then output as the roll angle deviation.

[0012] In some embodiments, the method further includes: Based on the preoperatively planned lesion location, a target navigation path is generated from the bronchial inlet to the lesion. The target navigation path contains multiple three-dimensional path points. Based on the target navigation path, calculate the theoretical yaw angle required at the tip of the duct at the bifurcation bend of the bronchus. During the surgery, when the robotic arm carrying the bronchoscope moves to the position corresponding to the theoretical swing angle, the posture of the robotic arm end is adjusted according to the roll angle deviation, and the lever drive mechanism on the clamping module is driven to automatically swing the front end of the catheter to the theoretical swing angle.

[0013] Secondly, the present invention also provides a control device for a bronchoscopy-assisted robot, the device comprising: The path acquisition module is used to acquire multiple Cartesian space path points that the end effector of the bronchoscope-assisted robot needs to pass through. The sequence conversion module is used to convert each path point into a sequence of joint angles for each joint of the robotic arm through inverse kinematics; The trajectory generation module is used to plan the trajectory of each joint in the joint space based on the joint angle sequence and using B-spline curves to generate the initial joint motion trajectory. The model solving module is used to establish an optimization mathematical model with the goal of minimizing total motion time and joint impact, and to solve the optimization mathematical model using an improved sparrow algorithm to obtain the optimal time series; The trajectory optimization module is used to redistribute the time of the initial joint motion trajectory using the optimal time series to obtain the optimized joint motion trajectory. The motion control module is used to control the robotic arm to move according to the optimized joint motion trajectory; The image acquisition module is used to acquire endoscopic images in real time through the bronchoscope on the bronchoscope-assisted robot, and to acquire the corresponding virtual navigation images generated in the electromagnetic navigation system. The image processing module is used to preprocess virtual navigation images and endoscopic images; The metric establishment module is used to establish a similarity metric for image registration based on the region grayscale values ​​of the preprocessed image and using a rigid transformation model. The rigid transformation model includes translation parameters, rotation parameters, and scaling parameters, and the similarity metric uses a structural similarity index. The deviation calculation module is used to perform global optimization of the similarity metric using an improved particle swarm optimization algorithm, and to solve for the roll angle deviation between the virtual navigation image and the endoscope image. The image registration module is used to register the virtual navigation image and the endoscope image based on the calculated roll angle deviation, so that the roll angle directions of the two are consistent.

[0014] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the control method of the bronchoscopy-assisted robot provided in the first aspect.

[0015] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the control method for the bronchoscopy-assisted robot provided in the first aspect.

[0016] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the control method for the bronchoscopy-assisted robot provided in the first aspect.

[0017] The beneficial effects of this invention are as follows: The control method for the bronchoscopy-assisted robot provided by this invention first plans the trajectory in joint space using B-spline curves and establishes an optimization mathematical model with the goal of minimizing total motion time and joint impact. An improved sparrow algorithm is used to solve for the optimal time series, and the initial trajectory is time-redistributed to obtain the optimized joint motion trajectory. This effectively suppresses vibration and impact during the start-up, shutdown, and movement of the robotic arm, significantly improving the spatial registration accuracy of the electromagnetic navigation system. Based on this, virtual navigation images and real-time endoscope images are further acquired. A similarity metric is established based on regional grayscale values ​​and a rigidity transformation model. An improved particle swarm optimization algorithm is used to globally optimize and solve for the roll angle deviation and perform registration, ensuring that the roll angle directions of the two images are consistent, providing an accurate attitude reference for subsequent precise catheter guidance. This method achieves high-precision, low-jitter control of the bronchoscopy-assisted robot through the synergy of trajectory optimization and image registration.

[0018] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a control method for a bronchoscopy-assisted robot according to an embodiment of the present invention. Figure 2 This is a schematic diagram comparing the location trajectories of three planning methods according to an embodiment of the present invention; Figure 3 This is a schematic diagram comparing the acceleration and jerk of three planning methods according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the optimal particle fitness curves of the JAKA robotic arm under four different algorithm planning methods, as shown in an embodiment of the present invention. Figure 5 This is a schematic diagram showing the registration error comparison according to an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the optimal particle fitness under five algorithms in one embodiment of the present invention; Figure 7 This is a schematic diagram of an electronic device structure provided in an embodiment of this application. Detailed Implementation

[0020] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] It should be noted that references to "an embodiment," "embodiment," "example embodiment," etc., in this specification refer to the described embodiment including specific features, structures, or characteristics; however, not every embodiment must include these specific features, structures, or characteristics. Furthermore, such expressions do not refer to the same embodiment. Moreover, when describing specific features, structures, or characteristics in conjunction with embodiments, whether or not explicitly described, it is indicated that incorporating such features, structures, or characteristics into other embodiments is within the knowledge of those skilled in the art.

[0022] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0023] In some embodiments, such as Figure 1 As shown, a control method for a bronchoscopy-assisted robot is provided, the specific method including: S101, obtain multiple Cartesian space path points that the end effector of the bronchoscope-assisted robot needs to pass through.

[0024] Specifically, the bronchoscopy-assisted robot includes a six-DOF robotic arm, a gripping module, and a linear delivery module, used to assist physicians in delivering the bronchoscope to the location of lung lesions. An electromagnetic navigation system (such as the Aurora series devices) generates a low-frequency three-dimensional magnetic field through a field generator and uses miniature electromagnetic sensors (five-DOF or six-DOF sensors) installed in the bronchoscope's working channel to acquire real-time spatial position and attitude information of the catheter tip, including coordinates (X, Y, Z) as well as pitch, yaw, and roll angles. This navigation system works in conjunction with the robot to provide real-time positioning feedback.

[0025] First, the multiple Cartesian space path points that the end effector of the bronchoscopy-assisted robot needs to traverse are obtained. These path points originate from the preoperative planning stage: the surgeon uses software such as 3DSlicer or Mimics to segment and 3D reconstruct the patient's lung CT images, marking the location of lesions and generating a navigation path from the bronchial inlet to the lesion. This path consists of a series of coordinate points in 3D space, i.e., Cartesian space path points. The end effector of the robot needs to pass through these points sequentially to complete the delivery.

[0026] S102 converts each path point into a sequence of joint angles for each joint of the robotic arm through inverse kinematics.

[0027] Specifically, inverse kinematics refers to the process of solving for the joint angles given the pose of the end effector in Cartesian space. In this embodiment, the inverse kinematics equations of the six-DOF manipulator are solved analytically to obtain multiple sets of joint angle solutions corresponding to each end effector pose, and the optimal solution with the smallest error is selected. The joint angle sequence is the angle values ​​(θ1 to θ6) of the six joints corresponding to each path point.

[0028] S103, based on the joint angle sequence, uses B-spline curves to plan the trajectory of each joint in the joint space and generate the initial joint motion trajectory.

[0029] Specifically, joint space trajectory planning refers to establishing a continuous curve that varies with time, using joint angles as a function. B-spline curves are parametric curves with good smoothness and local support, ensuring the continuity of joint angular displacement, angular velocity, and angular acceleration. This embodiment uses a seventh-order B-spline curve, whose mathematical representation can be expressed as: ; ; in, Represents intermediate process points; Represents curve parameters, t Represents time.

[0030] It has C 2 Continuity avoids abrupt acceleration changes, thereby reducing vibration and impact during robotic arm movement. B-spline interpolation is performed on the angle sequence of each joint to obtain a continuous function of the joint angles over time, i.e., the initial joint motion trajectory. Figure 2 and Figure 3 The effects of different planning methods were compared, showing that the acceleration and jerk curves of the seventh-order B-spline curve are smoother. The above process generates a smooth initial trajectory, providing a foundation for subsequent optimization, while reducing the impact on the robotic arm during start-up and shutdown, protecting the robotic arm and improving surgical stability.

[0031] S104. An optimization mathematical model is established with the goal of minimizing total motion time and joint impact. The improved sparrow algorithm is used to solve the optimization mathematical model to obtain the optimal time series.

[0032] Specifically, optimizing the mathematical model refers to representing the performance indicators (total time, impact) of the robotic arm during its movement using mathematical functions, and finding the optimal parameters while satisfying motion constraints. The movement time T between adjacent path points is considered. i As design variables, the limit values ​​of angular velocity, angular acceleration, and angular jerk of each joint of the robotic arm are used as constraints, and the fitness function is the weighted sum of the total motion time of all path segments and the impact cost function. ; Where t0 and t f These are the initial time and the termination time. and These are weighting coefficients. f1 is a time-related cost function (e.g., it may be related to the evolution rate of the system state, control energy consumption, and other time-related factors), and f2 is a shock-related cost function (e.g., it may be related to the magnitude of the shock and the losses caused by the shock). The normalization coefficient for the time term, Here, x(t) is the normalization coefficient for the impact term, x(t) is the motion state of the robotic arm at time t, u(t) is the control command / joint torque of the robotic arm, t is time, and I(t) is the joint jerk, i.e., the derivative of the acceleration.

[0033] The improved sparrow algorithm is an optimization of the standard sparrow search algorithm. It uses an elite back-learning strategy to generate the initial population to improve solution quality, and selects either Cauchy mutation or Gaussian mutation during iteration based on the population fitness variance (to escape local optima and accelerate convergence, respectively), thereby solving for the optimal time series (time allocation of each path segment) that minimizes the fitness function. Simulation experiments ( Figure 4 This demonstrates that the improved sparrow algorithm has a faster convergence speed and a stronger ability to escape local optima compared to the standard sparrow algorithm, genetic algorithm, and particle swarm optimization algorithm. The optimal time series obtained, from path point 1 to 8, has a total time of 39.4536 seconds. The beneficial effects of this step are: while satisfying the robotic arm motion constraints, it simultaneously optimizes the total surgical time and motion impact, making the robotic arm movement both fast and smooth. Experiments have verified that it can reduce the registration error by an average of 11.34%.

[0034] Optionally, improvements to the sparrow algorithm include: using an elite reverse learning strategy to generate an initial sparrow population; calculating the population fitness variance during the iteration process; performing Cauchy mutation on the current best sparrow individual when the fitness variance is lower than a preset threshold; and performing Gaussian mutation on the elite sparrow individuals when the fitness variance is higher than a preset threshold.

[0035] Specifically, elite reverse learning is a method to enhance the diversity of the initial solution set. Its core idea is: after randomly generating the initial population, select several individuals with better fitness as elite individuals, and then generate a reverse solution for each elite individual within its search space boundary. Let the d-th dimension of the search boundary be... If the elite individual's position is x, then the reverse solution is: ; In the formula, γ is any number between 0 and 1. Let d be the lower bound for the search in dimension d. The upper bound of the d-th dimension search For elite individualsi At the original position in the d-th dimension, For elite individuals i The inverse solution in the d-th dimension.

[0036] The reverse solution is merged with the original elite individuals, and the individuals with the best fitness are selected to form the initial population. This strategy ensures that the initial sparrows are more evenly distributed in the solution space, avoiding clustering caused by random initialization. This process improves the quality of the initial population, accelerates algorithm convergence, and thus obtains the optimal time series more quickly.

[0037] Population fitness variance reflects the dispersion of individual fitness in the current population. A larger value indicates greater individual diversity, suggesting the algorithm is in a global exploration phase; a smaller value indicates that individuals are becoming more uniform, and the algorithm may get stuck in a local optimum. Specifically, the variance is calculated by counting the fitness values ​​of all sparrow individuals. This variance serves as the basis for determining whether the algorithm has prematurely converged. The beneficial effect of this step is that it allows for real-time monitoring of the population state, providing a basis for subsequent adaptive mutation decisions.

[0038] Then, when the fitness variance falls below a preset threshold, the algorithm is considered to have entered premature convergence. At this point, a Cauchy mutation operation is performed on the current best sparrow individual. Cauchy mutation refers to generating random perturbations using a Cauchy distribution. The Cauchy distribution has a long tail, which can generate random numbers with a large step size, thereby helping the individual escape local optima. The mutation formula is: ; Where λ follows a Cauchy distribution. For the optimal individual, This is a mutated individual. A large perturbation is introduced when the algorithm might stall, enhancing global search capabilities and avoiding getting trapped in local optima.

[0039] On the other hand, when the fitness variance exceeds a preset threshold, it indicates that the population is still relatively dispersed. In this case, Gaussian mutation is performed on the elite sparrow individuals. Gaussian mutation refers to generating random perturbations using a Gaussian (normal) distribution. The energy of the Gaussian distribution is concentrated near the mean, and the small step size perturbations generated are beneficial for fine-grained searching near the current optimal solution. The above process enhances the local exploitation capability during normal algorithm iteration, accelerating convergence to the global optimum.

[0040] Through the aforementioned adaptive mutation strategy, the improved sparrow algorithm can dynamically balance global exploration and local exploitation, achieving faster convergence speed and higher solution accuracy compared to the standard sparrow algorithm. Figure 4 Simulation experiments show that the improved sparrow algorithm can converge to a better fitness value within 50 iterations, which is better than the standard sparrow algorithm, genetic algorithm and standard particle swarm algorithm.

[0041] Optionally, an optimization mathematical model is established with the goal of minimizing total motion time and joint impact, including: using the motion time between adjacent path points as a design variable; using the limit values ​​of angular velocity, angular acceleration, and angular jerk of each joint of the robotic arm as constraints; and using the weighted sum of the total motion time of all path segments and the impact cost function as the fitness function.

[0042] Specifically, firstly, the movement time between adjacent path points is used as a design variable. Given n path points, there are n-1 path segments between adjacent points, and the movement time for each path segment is denoted as T. i (i=1,2,...,n-1). These time variables are parameters to be optimized, and their values ​​determine the total time required for the robotic arm to complete the entire trajectory, as well as the rate of change of velocity and acceleration in each segment. For example, 8 path points correspond to 7 time intervals, and the total time is T for each interval. i The sum of these. The beneficial effect of this step is that it transforms the continuous time allocation problem into discrete optimization variables, which facilitates mathematical modeling and algorithmic solutions.

[0043] Secondly, the limits of angular velocity, angular acceleration, and angular jerk for each joint of the robotic arm are used as constraints. Angular velocity is the speed of joint rotation, angular acceleration is the rate of change of velocity, and angular jerk (also known as acceleration) is the rate of change of acceleration. Excessive angular velocity or angular acceleration will cause the robotic arm to move too violently, generating impact; excessive angular jerk will cause severe vibration, affecting surgical precision. In this embodiment, based on the performance parameters of the JAKA MiniCobo robotic arm, maximum values ​​are set for each joint: joint 1 angular velocity 100° / s, angular acceleration 60° / s. 2 Angular acceleration 60° / s 3 Joint 2 angular velocity 95° / s, angular acceleration 60° / s 2 Angular acceleration 60° / s 3 Joint 3: angular velocity 100° / s, angular acceleration 75° / s 2 Angular acceleration 55° / s 3 During the optimization process, any candidate time allocation scheme must ensure that the actual motion parameters of each joint do not exceed these limits; otherwise, it is an infeasible solution. This step ensures that the optimized trajectory meets the physical performance and safety requirements of the robotic arm, avoiding motor overload or loss of motion control.

[0044] Finally, the fitness function is the weighted sum of the total motion time for all path segments and the impact cost function. The total motion time is ΣT. i This reflects surgical efficiency; shorter time is better. The impact cost function quantifies the vibration and impact during robotic arm movement and is typically related to the integral of the square of the angular acceleration over time. Weighting coefficients are introduced... and The two optimization objectives (time optimum and impact minimum) are combined into a single objective function: J = (Total Time) + • (Impact Costs). The weighting coefficient can be adjusted according to actual needs; for example, it can be increased when surgical efficiency is emphasized. When stability is a priority, it can be increased In this embodiment, the constraint amplification factor was set to 2.8 for each experiment. The beneficial effect of this step is that it achieves a unified representation of multi-objective optimization, enabling the improved sparrow algorithm to simultaneously consider surgical speed and motion stability, resulting in a trajectory that is both fast and low-impact.

[0045] S105, the initial joint motion trajectory is redistributed in time using the optimal time series to obtain the optimized joint motion trajectory.

[0046] Specifically, time reallocation refers to readjusting the time nodes in the B-spline curve according to the optimal time sequence, thereby changing the timing of the robotic arm passing through each path point without altering the geometry of the path. Specifically, the uniform or preset time parameters in the initial trajectory are replaced with the optimized time increments for each segment, and the control points and node vectors of the B-spline curve are recalculated to generate a new joint motion trajectory that satisfies the optimal time allocation. This optimized trajectory ensures that the robotic arm's velocity, acceleration, and jerk at each joint do not exceed allowable ranges, while minimizing total time and impact. The optimized curves showing the position, velocity, acceleration, and jerk of each joint over time exhibit smooth transitions. The beneficial effect of this step is that it transforms abstract optimized values ​​into actual executable robotic arm motion commands, ensuring both high efficiency and stability during surgery.

[0047] S106 controls the robotic arm to move according to the optimized joint motion trajectory.

[0048] Specifically, the robotic arm is controlled via a motion control card (such as the Advantech PCI-1288). This card receives pulse commands from the host computer and drives the motors to operate at the planned speed and acceleration, thus enabling the robotic arm's end effector to move along a predetermined trajectory. During the movement, because the trajectory is optimized, the robotic arm's start-up, stopping, and joint speed changes are smooth and shock-free, effectively reducing data jitter in the electromagnetic navigation sensors caused by mechanical vibration. Figure 5 The registration accuracy experiment shown demonstrates that using a robotic arm with the optimized trajectory for model registration reduced the average registration error from 3.4810 mm to 3.0865 mm, a reduction of 11.34%. The beneficial effect of this step is that actually executing the optimized trajectory significantly improves the spatial registration accuracy of the electromagnetic navigation system, providing a more accurate coordinate reference for subsequent image registration and lesion localization.

[0049] S107 uses a bronchoscope on a bronchoscope-assisted robot to acquire endoscopic images in real time and obtain corresponding virtual navigation images generated in the electromagnetic navigation system.

[0050] Specifically, endoscopic images refer to real-time images of the airway interior captured by a miniature camera installed at the tip of the bronchoscope, reflecting the actual anatomical structure. Virtual navigation images are virtual perspective images generated in electromagnetic navigation software based on a three-dimensional lung model reconstructed from the patient's preoperative CT data, according to the real-time position of sensors. Both are displayed simultaneously on the monitor during the procedure. However, because electromagnetic sensors are typically five degrees of freedom (lacking roll angle information), there is an angular deviation between the virtual navigation images and the endoscopic images in the direction of rotation around the catheter axis. In this embodiment, endoscopic images are acquired using an Olympus BF-XP190 bronchoscope and imaging system, while virtual navigation images are generated by navigation software independently developed in the laboratory.

[0051] S108, preprocess the virtual navigation image and endoscope image.

[0052] Optionally, the virtual navigation image and the endoscope image are preprocessed, including converting the virtual navigation image and the endoscope image into grayscale images, and performing color inversion processing on the grayscale virtual navigation image.

[0053] Specifically, preprocessing includes converting the color image to a grayscale image and inverting the colors of the grayscale virtual navigation image. Grayscale conversion uses a weighted average method, with the following formula: ; gray represents the pixel value of a grayscale image, indicating brightness. R, G, and B represent the pixel values ​​of the red, green, and blue channels in a color image, respectively.

[0054] Convert the RGB three channels to a single channel luminance value. Inversion is performed on each pixel value in the grayscale image. ; Where inverted_pixel is the pixel value of the grayscale image after inversion, and original_pixel is the pixel value of the original grayscale image.

[0055] The dark areas are brightened, and the bright areas are darkened. Because the camera lighting settings in the virtual navigation image result in indistinct features, color inversion enhances the contrast of structures such as blood vessels and bronchial walls, facilitating subsequent registration. This process eliminates differences in illumination and color, enhances image features, and improves the accuracy and robustness of registration.

[0056] S109. Based on the regional grayscale values ​​of the preprocessed image, a rigid transformation model is used to establish a similarity measure for image registration.

[0057] The rigid transformation model includes translation parameters, rotation parameters, and scaling parameters, and the similarity measure uses the structural similarity index.

[0058] Specifically, rigid transformations refer to geometric transformations that only involve rotation, translation, and uniform scaling, without changing the shape and angular relationships of the image. They are suitable for global pose deviations caused by camera viewpoint and sensor mounting in this scenario. The Structural Similarity Index (SSIM) is a metric for evaluating the structural similarity between two images. ; Its main parameters include reference image r With the registered image f grayscale mean and Gray-scale variance and and covariance The formula introduces a stability constant. a and b This is to avoid numerical instability caused by the denominator approaching zero.

[0059] It comprehensively considers three dimensions: brightness, contrast, and structure; the closer the value is to 1, the more similar the images are. In this embodiment, the virtual navigation image is used as the reference image, and the endoscope image is used as the image to be registered. By adjusting the rigid transformation parameters (scaling ratio, rotation angle, horizontal offset, and vertical offset), the SSIM value between the transformed image and the reference image is calculated as a similarity measure. The beneficial effect of this step is that it establishes a quantifiable registration objective function, providing optimization direction for subsequent optimization algorithms.

[0060] S110 utilizes an improved particle swarm optimization algorithm to globally optimize the similarity metric, thereby solving for the roll angle deviation between the virtual navigation image and the endoscope image.

[0061] Specifically, the improved particle swarm optimization (PSO) algorithm introduces differential mutation and dynamic learning factor strategies on top of the standard PSO algorithm. The optimization process includes: initializing the particle swarm, where each particle's position vector contains four dimensions—image scaling, rotation angle, horizontal offset, and vertical offset; calculating the fitness value of each particle based on a similarity metric (i.e., SSIM value); updating the individual historical best position of each particle and the global historical best position of the particle swarm; and dynamically adjusting the learning factor during iteration, so that the individual learning factor c1 decreases with increasing iteration count, while the social learning factor c2 increases with increasing iteration count. ; Among them, c11 and c 21 Let c be a constant; in this paper, we take c. 11 =c 21 =1, where n represents the current iteration number and represents the maximum number of generations.

[0062] A differential mutation strategy is introduced to generate a mutation vector, and the target vector and the mutation vector are cross-operated to generate a test vector. The particle position is updated according to the test vector, and the iteration is repeated until the termination condition is met. The global optimal position is output as the roll angle deviation. Figure 6 Experimental results show that the improved particle swarm optimization algorithm outperforms the standard PSO, DE, GA, and SA algorithms in multiple evaluation metrics, including MSE, PSNR, and CC. MSE is reduced by 22.48%, PSNR is increased by 2.22%, and CC is increased by 1.87%. The calculated roll angle deviation is an angular value, which can be used to calculate the required rotation angle of the virtual navigation image relative to the endoscopic image through registration. The beneficial effects of this step are: rapid and accurate calculation of the roll angle deviation, solving the problem of missing roll angle information in five-DOF sensors, and providing accurate parameters for subsequent catheter attitude adjustment.

[0063] Optionally, an improved particle swarm optimization algorithm is used to globally optimize the similarity metric to solve for the roll angle deviation between the virtual navigation image and the endoscopic image. This includes: initializing the particle swarm, where each particle's position vector contains the image scaling ratio, rotation angle, horizontal offset, and vertical offset; calculating the fitness value of each particle based on the similarity metric; updating the individual historical best position of each particle and the global historical best position of the particle swarm; dynamically adjusting the learning factor during iteration, so that the individual learning factor decreases with the number of iterations, while the social learning factor increases with the number of iterations; introducing a differential mutation strategy to generate a mutation vector, and performing a cross operation between the target vector and the mutation vector to generate a trial vector; updating the particle position based on the trial vector, repeating the iteration until the termination condition is met, and outputting the global best position as the roll angle deviation.

[0064] Specifically, firstly, the particle swarm is initialized. Each particle's position vector includes the image scaling ratio, rotation angle, horizontal offset, and vertical offset. Particle swarm optimization (PSO) is a swarm intelligence algorithm where each particle represents a candidate solution. In this embodiment, each particle's position is a four-dimensional vector X=(s,θ,dx,dy), where s represents the image scaling ratio (typically ranging from 0.8 to 1.2), θ represents the rotation angle (i.e., the roll angle deviation to be solved, ranging from -180° to 180°), dx represents the horizontal translation offset (in pixels), and dy represents the vertical translation offset. The particle swarm size is set to 50, and the positions and velocities of all particles are randomly initialized within their respective value ranges. The beneficial effect of this step is that it defines the search dimension of the solution space, providing an initial population for subsequent iterations.

[0065] Next, the fitness value of each particle is calculated based on a similarity metric. The similarity metric used is the Structural Similarity Index (SSIM), whose value ranges from -1 to 1, with values ​​closer to 1 indicating greater similarity between the two images. The transformation parameters (scaling, rotation, translation) represented by each particle are applied to the endoscope image to be registered, resulting in a transformed image. The SSIM value between this transformed image and the reference virtual navigation image is then calculated, and this SSIM value is used as the particle's fitness. A higher fitness (closer to 1) indicates more accurate transformation parameters. The beneficial effect of this step is that it transforms the registration problem into an optimization problem of maximizing SSIM, providing an evaluation basis for the particle swarm optimization algorithm.

[0066] Then, update the individual historical best position of each particle and the global historical best position of the particle swarm. The individual historical best position refers to the position vector corresponding to the highest fitness achieved by the particle in all iterations; the global historical best position refers to the position vector corresponding to the highest fitness of all particles in the entire particle swarm in all iterations. After calculating the fitness of all particles in each iteration, compare the current fitness of each particle with its individual historical best fitness; if it is better, replace it; then compare the current best fitness of all particles with the global historical best; if it is better, update the global best. The beneficial effect of this step is: recording the best candidate solution during the search process and guiding particles to move towards a better region.

[0067] Next, during the iteration process, the learning factors are dynamically adjusted so that the individual learning factor decreases with increasing iteration count, while the social learning factor increases. Learning factors c1 and c2 control the weights of a particle's flight towards its own historical best position and the global historical best position, respectively. In standard PSO, c1 and c2 are usually fixed values ​​(e.g., 2), which can easily lead to premature convergence or slow convergence. This embodiment employs a dynamic adjustment strategy: ; Where n is the current iteration number and N is the maximum iteration number. In the early stages of iteration, c1 is larger, and particles rely more on their own experience to explore extensively; in the later stages of iteration, c2 is larger, and particles gather more towards the optimal position in the swarm, accelerating convergence. The beneficial effect of this step is: balancing global search and local exploration capabilities, improving convergence accuracy and speed.

[0068] Then, a differential mutation strategy is introduced to generate mutation vectors, and the target vector is crossed with the mutation vectors to generate trial vectors. Differential mutation is the core operation of the differential evolution algorithm, used to enhance population diversity. Specifically, three distinct individuals X are randomly selected from the current population. r1,g X r2,g X r3,g According to the formula: Generate mutation vectors V ,in F This is the scaling factor (values ​​range from 0.5 to 1.0).

[0069] After the mutation operation, the target vector needs to be... With the mutation vector The final test vector is generated by performing binomial crossover. Perform the crossover operation according to the following formula: ; in, It is an integer randomly selected from the set {1,2,...,D} to ensure the mutation vector At least one dimension of information is retained, cross probability CR It is a constant within the interval (0,1).

[0070] Finally, the particle position is updated based on the trial vector, and the iteration is repeated until the termination condition is met. The globally optimal position is output as the roll angle deviation. Specifically, the fitness of the trial vector U is calculated. If the fitness of U is better than that of the current particle X, then... i If the fitness is such that X is replaced by U, then... i Otherwise, keep X. i Then, the process proceeds to the next iteration, repeating operations such as calculating fitness, updating individual and global optima, dynamically adjusting learning factors, and differential mutation crossover. Termination conditions can include reaching the maximum number of iterations (e.g., 50) or achieving no significant improvement in global optima fitness across multiple generations. After the iteration ends, the rotation angle θ in the global optima position vector is output, representing the roll angle deviation between the virtual navigation image and the endoscopic image.

[0071] S111, based on the solved roll angle deviation, register the virtual navigation image and the endoscope image to make their roll angle directions consistent.

[0072] Specifically, registration refers to rotating the endoscopic image according to the calculated roll angle deviation, or rotating the viewpoint of the virtual navigation image by a corresponding angle, so that the roll angle directions displayed on the screen are exactly the same for both. Specifically, in the navigation software, the roll angle parameter of the virtual camera is increased or decreased by this deviation value, thereby aligning the bronchial bifurcation direction in the virtual navigation image with the bifurcation direction in the real-time endoscopic image. Before registration, there is a significant angular difference between the two; after registration, they are completely identical. At this point, the two images seen by the doctor on the monitor have the same spatial orientation, making it easier to determine which branch the catheter should turn to. This step eliminates the roll angle error between the navigation image and the endoscopic image, providing the doctor with intuitive and consistent visual guidance, and providing an accurate posture reference for subsequent assisted flexion segment navigation functions.

[0073] Optionally, after registering the virtual navigation image and the endoscopic image, the above method further includes: generating a target navigation path from the bronchial inlet to the lesion based on the preoperatively planned lesion location, the target navigation path containing multiple three-dimensional path points; calculating the theoretical sway angle required at the tip of the catheter at the bifurcation bend of the bronchus according to the target navigation path; during the operation, when the robotic arm carrying the bronchoscope moves to the position point corresponding to the theoretical sway angle, adjusting the posture of the robotic arm end based on the roll angle deviation, and driving the lever drive mechanism on the clamping module to automatically sway the tip of the catheter to the theoretical sway angle.

[0074] Specifically, firstly, based on the preoperatively planned lesion location, a target navigation path is generated from the bronchial inlet to the lesion. This target navigation path contains multiple three-dimensional path points. During the preoperative planning phase, the physician uses 3DSlicer or Mimics software to segment and 3D reconstruct the patient's lung CT images, marking the lesion location on the reconstructed model. The navigation software automatically generates an optimized path based on the lesion location, starting from the main bronchial inlet, passing through various bronchial branches, and finally reaching the lesion. This path consists of a series of dense 3D coordinate points, each containing X, Y, and Z coordinates. These path points are used not only for the robotic arm's trajectory planning but also for identifying the locations of bronchial bifurcation bends.

[0075] Secondly, based on the target navigation path, the theoretical yaw angle required for the catheter tip at the bronchial bifurcation bend is calculated. In the path data, the bronchial bifurcation point represents a significant change in path direction. By performing differential calculations on the direction vectors between adjacent path points, the bend can be identified, and the required bend angle of the catheter can be determined. Specifically, a direction vector V1 is taken before the bifurcation point, and a direction vector V2 is taken after the bifurcation point. The angle between V1 and V2 is calculated; this angle is the theoretical yaw angle required for the catheter tip. The path data is processed using a C++ program, and mathematical calculations are performed using the Eigen library to calculate the angle of each bend and mark the theoretical bend points.

[0076] Finally, during the surgery, when the robotic arm carrying the bronchoscope reaches the position corresponding to the theoretical yaw angle, the attitude of the robotic arm's end effector is adjusted according to the roll angle deviation, and the lever drive mechanism on the clamping module is driven to automatically yaw the catheter tip to the theoretical yaw angle. Since the electromagnetic sensor lacks roll angle information, directly driving the lever according to the theoretical yaw angle may result in an inconsistency between the actual yaw direction and the direction indicated by the virtual navigation. Therefore, it is necessary to first rotate the robotic arm's end flange (i.e., rotate it around the catheter axis) according to the roll angle deviation to align the actual reference frame of the catheter tip with the reference frame of the virtual navigation image, and then drive the lever to yaw the catheter tip by the theoretical angle. The lever drive mechanism on the clamping module consists of a motor and a linkage mechanism. After receiving the control command, the motor rotates, driving the lever head to move the bronchoscope's operating lever through the linkage, thereby pulling the steel wire to bend the catheter tip. Experiments show that after adopting this auxiliary function, the average operation time for inexperienced operators decreased from 63.375 seconds to 51.75 seconds, a reduction of 18.34%. The beneficial effects of this step are: to achieve automatic assisted bending of the curved section at the bifurcation point, significantly reducing the difficulty of operation for doctors and the operation time, and improving the efficiency and success rate of the operation.

[0077] Based on the same inventive concept, this application also provides a control device for a bronchoscopy-assisted robot to implement the control method of the bronchoscopy-assisted robot described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the control device for a bronchoscopy-assisted robot provided below can be found in the limitations of the control method for the bronchoscopy-assisted robot described above, and will not be repeated here.

[0078] In one embodiment, a control device for a bronchoscopy-assisted robot is provided, the device comprising: The path acquisition module is used to acquire multiple Cartesian space path points that the end effector of the bronchoscope-assisted robot needs to pass through. The sequence conversion module is used to convert each path point into a sequence of joint angles for each joint of the robotic arm through inverse kinematics; The trajectory generation module is used to plan the trajectory of each joint in the joint space based on the joint angle sequence and using B-spline curves to generate the initial joint motion trajectory. The model solving module is used to establish an optimization mathematical model with the goal of minimizing total motion time and joint impact, and to solve the optimization mathematical model using an improved sparrow algorithm to obtain the optimal time series; The trajectory optimization module is used to redistribute the time of the initial joint motion trajectory using the optimal time series to obtain the optimized joint motion trajectory. The motion control module is used to control the robotic arm to move according to the optimized joint motion trajectory; The image acquisition module is used to acquire endoscopic images in real time through the bronchoscope on the bronchoscope-assisted robot, and to acquire the corresponding virtual navigation images generated in the electromagnetic navigation system. The image processing module is used to preprocess virtual navigation images and endoscopic images; The metric establishment module is used to establish a similarity metric for image registration based on the region grayscale values ​​of the preprocessed image and using a rigid transformation model. The rigid transformation model includes translation parameters, rotation parameters, and scaling parameters, and the similarity metric uses a structural similarity index. The deviation calculation module is used to perform global optimization of the similarity metric using an improved particle swarm optimization algorithm, and to solve for the roll angle deviation between the virtual navigation image and the endoscope image. The image registration module is used to register the virtual navigation image and the endoscope image based on the calculated roll angle deviation, so that the roll angle directions of the two are consistent.

[0079] This application also provides an electronic device, in some embodiments, referring to... Figure 7 As shown, the electronic device 700 includes an input unit 710, a memory 720, a processor 730, and an output unit 740. The memory 720 stores program instructions that can be executed on the processor 730. The processor 730 can execute the control method and / or technical solution of the bronchoscopy-assisted robot based on the foregoing embodiments by calling the program instructions. The electronic device 700 can be a mobile terminal device such as a mobile phone or computer.

[0080] Furthermore, embodiments of this application also provide a computer-readable storage medium for storing a computer program for executing a control method for a bronchoscopy-assisted robot. For example, computer program instructions, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. The program instructions for invoking the methods of this application may be stored in a fixed or removable storage medium, and / or transmitted via data streams in broadcast or other signal carrying media, and / or stored in a storage medium that operates according to the program instructions.

[0081] Obviously, those skilled in the art should understand that the modules or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0082] The technical features of the above embodiments can be arbitrarily integrated. For the sake of brevity, not all possible integrations of the technical features in the above embodiments are described. However, as long as the integration of these technical features does not contradict each other, they should be considered to be within the scope of this specification.

[0083] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A control method for a bronchoscopy-assisted robot, characterized in that, The method includes: Obtain multiple Cartesian space path points that the end effector of the bronchoscope-assisted robot needs to pass through; The path points are converted into a sequence of joint angles for each joint of the robotic arm using inverse kinematics. Based on the joint angle sequence, B-spline curves are used to plan the trajectory of each joint in the joint space to generate the initial joint motion trajectory. An optimization mathematical model is established with the goal of minimizing total motion time and joint impact. The improved sparrow algorithm is used to solve the optimization mathematical model to obtain the optimal time series. The initial joint motion trajectory is time-redistributed using the optimal time series to obtain an optimized joint motion trajectory. Control the robotic arm to move according to the optimized joint motion trajectory; The bronchoscope on the bronchoscope-assisted robot acquires endoscopic images in real time and obtains corresponding virtual navigation images generated in the electromagnetic navigation system. The virtual navigation image and endoscope image are preprocessed; Based on the regional grayscale values ​​of the preprocessed image, a rigid transformation model is used to establish a similarity metric for image registration; the rigid transformation model includes translation parameters, rotation parameters, and scaling parameters, and the similarity metric uses a structural similarity index; An improved particle swarm optimization algorithm is used to globally optimize the similarity metric, and the roll angle deviation between the virtual navigation image and the endoscope image is solved. Based on the calculated roll angle deviation, the virtual navigation image and the endoscope image are registered to ensure that their roll angle directions are consistent.

2. The control method for the bronchoscopy-assisted robot as described in claim 1, characterized in that, The improvements to the Sparrow Algorithm include: An elite reverse learning strategy was used to generate an initial sparrow population. During the iteration process, the population fitness variance is calculated; When the fitness variance is lower than a preset threshold, Cauchy mutation is performed on the current best sparrow individual; When the fitness variance is higher than a preset threshold, Gaussian mutation is performed on the elite sparrow individuals.

3. The control method for the bronchoscopy-assisted robot as described in claim 2, characterized in that, An optimization mathematical model is established with the objective of minimizing total motion time and joint impact, including: Use the travel time between adjacent path points as a design variable; The limits of angular velocity, angular acceleration, and angular jerk of each joint of the robotic arm are used as constraints. The fitness function is the weighted sum of the total motion time of all path segments and the impact cost function.

4. The control method for the bronchoscopy-assisted robot as described in claim 1, characterized in that, Preprocessing of the virtual navigation image and endoscope image includes: The virtual navigation image and endoscope image are converted into grayscale images, and the grayscale virtual navigation image is then inverted.

5. The control method for the bronchoscopy-assisted robot as described in claim 1, characterized in that, The similarity metric is globally optimized using an improved particle swarm optimization algorithm to solve for the roll angle deviation between the virtual navigation image and the endoscopic image, including: Initialize the particle swarm, where the position vector of each particle includes the image scaling ratio, rotation angle, horizontal offset, and vertical offset; Calculate the fitness value for each particle, the fitness value being based on the similarity metric; Update the individual historical best position of each particle and the global historical best position of the particle swarm; During the iteration process, the learning factor is dynamically adjusted so that the individual learning factor decreases as the number of iterations increases, while the social learning factor increases as the number of iterations increases. A differential mutation strategy is introduced to generate mutation vectors, and the target vector and mutation vectors are cross-operated to generate experimental vectors; The particle position is updated based on the experimental vector, and the iteration is repeated until the termination condition is met. The globally optimal position is then output as the roll angle deviation.

6. The control method for the bronchoscopy-assisted robot as described in any one of claims 1-5, characterized in that, The method further includes: Based on the preoperatively planned lesion location, a target navigation path is generated from the bronchial inlet to the lesion, and the target navigation path contains multiple three-dimensional path points; Based on the target navigation path, calculate the theoretical yaw angle required at the tip of the duct at the bifurcation bend of the bronchus. During the surgery, when the robotic arm carrying the bronchoscope moves to the position point corresponding to the theoretical swing angle, the posture of the robotic arm end is adjusted according to the roll angle deviation, and the lever drive mechanism on the clamping module is driven to automatically swing the front end of the catheter to the theoretical swing angle.

7. A control device for a bronchoscopy-assisted robot, characterized in that, The device includes: The path acquisition module is used to acquire multiple Cartesian space path points that the end effector of the bronchoscope-assisted robot needs to pass through. The sequence conversion module is used to convert each path point into a joint angle sequence of each joint of the robotic arm through inverse kinematics; The trajectory generation module is used to plan the trajectory of each joint in the joint space based on the joint angle sequence and using B-spline curves to generate the initial joint motion trajectory. The model solving module is used to establish an optimization mathematical model with the goal of minimizing total motion time and joint impact, and to solve the optimization mathematical model using an improved sparrow algorithm to obtain the optimal time series; The trajectory optimization module is used to redistribute the time of the initial joint motion trajectory using the optimal time series to obtain the optimized joint motion trajectory. The motion control module is used to control the robotic arm to move according to the optimized joint motion trajectory. The image acquisition module is used to acquire endoscopic images in real time through the bronchoscope on the bronchoscope-assisted robot, and to acquire the corresponding virtual navigation images generated in the electromagnetic navigation system. The image processing module is used to preprocess the virtual navigation image and the endoscope image; The metric establishment module is used to establish a similarity metric for image registration based on the regional grayscale values ​​of the preprocessed image and using a rigid transformation model; the rigid transformation model includes translation parameters, rotation parameters, and scaling parameters, and the similarity metric adopts a structural similarity index; The deviation calculation module is used to perform global optimization of the similarity metric using an improved particle swarm optimization algorithm to solve for the roll angle deviation between the virtual navigation image and the endoscope image. The image registration module is used to register the virtual navigation image and the endoscope image based on the calculated roll angle deviation, so that the roll angle directions of the two are consistent.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the control method for the bronchoscopic-assisted robot as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the control method for the bronchoscopy-assisted robot according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the control method for the bronchoscopic-assisted robot as described in any one of claims 1 to 6.

Images