An intelligently controlled six-link fracture reduction robot

Abstract

The present invention discloses an intelligently controlled six-link fracture reduction robot, comprising two fixed rings for mounting Kirschner wires, six connecting arms that are hingedly arranged between the two fixed rings and are uniformly distributed along the circumference, any two adjacent connecting arms are brought together at one end and separated at the other end in an eight-shaped shape, each connecting arm is provided with a telescopic device, and further comprises a control center (PC end), which is communicatively connected to the telescopic device and implements telescopic control, and is provided with a control module for controlling the relative position path of the two fixed rings and a path planning module for designing the control path, characterized in that the path planning module has an improved RRT* path planning algorithm. The present application solution can assist doctors in their operations, so as to better avoid accidental injuries through obstacle avoidance strategies, and better improve the reliability of the robot in the fracture rehabilitation assistance process.

Description

Technical Field

[0001] The present invention relates to the technical field of fracture reduction robots, and in particular to an intelligently controlled six-link fracture reduction robot. Background Art

[0002] During the rehabilitation process of fracture patients, especially limb fracture patients, it is usually necessary to use an orthopedic external fixator to fix the fracture site to avoid secondary damage to the fracture site during the rehabilitation process and accelerate the growth and recovery of the fracture site.

[0003] Existing bone external fixation devices can be divided into three different types according to their basic components and structures, namely unilateral / bilateral external fixators, circular external fixators (CEFs) and hybrid external fixators. In the treatment of long bone deformities of the lower limbs, the optimal use condition of external fixators is to allow micro-movement of the bone segments in the long bone axis while maintaining high stiffness in torsion, shear and bending. Although unilateral / bilateral external fixators can provide higher rigidity and early load-bearing capacity, they limit the axial micro-movement of the bone injury site and may therefore delay the healing process. In addition, asymmetric compression of the bone injury site is often observed in unilateral / bilateral external fixators, which will have an adverse effect on healing. Circular external fixators just eliminate these two problems. In circular external fixators, suitable Kirschner wires can be used to control axial micro-movement. In cases where the axial load on the bone increases, suitable Kirschner wires can be used in a targeted manner to eliminate excessive movement of the bone injury site. Furthermore, the high stiffness exhibited by circular external fixators in torsion, shear, and bending facilitates stability during bone correction. The multiplanar fixation provided by circular external fixators allows for the treatment of more complex bone injuries than unilateral or bilateral external fixators. For these reasons, circular external fixators have gained widespread popularity in orthopedics. Due to their biomechanically favorable stiffness properties, circular external fixators (CEFs) have been widely used to treat a variety of orthopedic conditions, particularly fracture reduction.

[0004] The structure of the existing conventional circular external fixator (CEF) type fracture reduction robot includes two fixing rings for installing Kirschner wires, and six connecting arms that are evenly distributed along the circumference are hingedly arranged between the two fixing rings. Any two adjacent connecting arms are close together at one end and separated at the other end to form an eight-shaped shape. Each connecting arm is provided with a telescopic device to enable it to have a telescopic function. When it is used for fracture rehabilitation, the two fixing rings are fixed to the two ends of the limb fracture through Kirschner wires to keep the limb fixed. When this fracture reduction robot is used, it can rely on the control of each telescopic arm to achieve adjustment and control of the relative position between the two fixing rings. The adjustment of the relative position between the two fixing rings will show different changes in relative position paths as the entire rehabilitation process of the fracture occurs.

[0005] Therefore, preoperative path planning technology is an important technology in fracture reduction surgery. Path information is obtained through programs and computer equipment, providing surgeons with more accurate path guidance, thereby improving surgical safety. Existing fracture reduction robots are usually designed to plan paths during fracture rehabilitation using the A* algorithm and the RRT* algorithm. The A* algorithm has a long calculation time and high memory consumption, and needs further improvement. In addition, the gap between the generated path and obstacles is very small, which will make it impossible for the robot to effectively avoid obstacles. In contrast, the RRT* algorithm can be effectively applied to path planning in complex environments, showing stronger search capabilities, and is therefore relatively more widely used.

[0006] However, during surgery, the complexity of the tissue surrounding the fractured bones makes it difficult for doctors to accurately determine their position. Consequently, collisions between bones can easily lead to secondary fractures and soft tissue damage. This requires ensuring collision-free reduction paths and quickly finding the optimal path. This places stricter demands on the algorithm. Existing RRT* algorithms still lack comprehensive consideration of the fracture reduction path, resulting in poor search efficiency and low collision detection efficiency, making it difficult to obtain the optimal reduction path.

[0007] Therefore, for those skilled in the art, how to provide an algorithm technology that can better help avoid important parts and avoid surgical injuries through obstacle avoidance strategies has become a problem to be considered and solved. Summary of the Invention

[0008] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is: how to provide an intelligently controlled six-link fracture reduction robot that can better avoid important parts through obstacle avoidance strategies and avoid surgical injuries.

[0009] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0010] An intelligently controlled six-link fracture reduction robot comprises two fixed rings for mounting Kirschner wires, six connecting arms being hingedly arranged between the two fixed rings and uniformly distributed along the circumference, any two adjacent connecting arms being brought together at one end and separated at the other end in a figure eight shape, each connecting arm being provided with a telescopic device, and a control center (PC end) being communicatively connected to the telescopic device and implementing telescopic control, the control center being provided with a control module for controlling the relative position path of the two fixed rings and a path planning module for designing the control path, characterized in that the path planning module has an improved RRT* path planning algorithm; the improved RRT* path planning algorithm comprising the following steps:

[0011] Step S1: Initialize the map and set the force field value and other related parameters;

[0012] Step S2: Introduce the artificial potential field method to generate random sampling points q new ;

[0013] Step S3: Using the cylindrical envelope method, q new Build a cylinder and check whether a collision occurs; if a collision occurs, return to step S2 and reselect the sample; if no collision occurs, continue to step S4;

[0014] Step S4: Find the tree node closest to the random sampling point q n , and use this node as q new The parent node of

[0015] Step S5: Find q new The minimum total path and reconnect;

[0016] Step S6: Determination q new arrive Q goal Is the distance less than the step size? If not, return to step 2 for sampling; if so, continue to step S7;

[0017] Step S7: Using Gaussian smoothing algorithm for fitting processing, adding the processed points to the new smoothed path set;

[0018] Step S8: Generate the final smooth planning path and the path planning task ends.

[0019] Thus, this solution utilizes an improved path planning algorithm within the path planning module of a conventional intelligently controlled six-link fracture reduction robot. The proposed improved algorithm performs path planning on the PC side, and the generated path node pose information is then imported into the control module, which then controls the motor rotation to achieve controlled reduction.

[0020] Furthermore, the step S1 includes the following steps: 1) initializing the map, setting the map size to 2×2×2 cm, setting the starting point to (-1,-1,-1), and setting the end point to (1,1,1); 2) setting the radius and height of the cylindrical obstacle to 0.05 and 0.2, respectively, and the position information to (0, 0, -0.1), (-0.6, -0.6, 0.4), (0.6, -0.6, 0.4), (-0.6, 0.6, 0.4), (0.6, 0.6, 0.4), (-0.6, -0.6, -0.4), (0.6, 0.6, 0.4), (-0.6, -0.6, -0.4) (0.6, 0.6, -0.4) -0.4); 3) Set the parameters of the artificial force field: the relative strength of attraction is 1, the relative strength of repulsion is 2, the repulsive influence range is 0.3, and the step size is 0.04.

[0021] This is because the above-mentioned environment, starting and ending point and obstacle setting data are based on a comprehensive consideration of the fracture environment characteristics and actual task requirements. At the same time, the advantages of the potential field parameter settings are reflected: the relative strength of gravity is set to 1, indicating that the gravitational field has a moderate impact on path planning, which can guide the path toward the end point, but not be too strong to cause the path to be too direct and ignore obstacles. The relative strength of repulsion is set to 2, which is higher than the relative strength of gravity, indicating that obstacle avoidance is the primary consideration in path planning, which helps to ensure the safety of the path. The repulsion influence range is set to 0.3. This range should be large enough to form a safe buffer zone around obstacles, but not so large that the path is too circuitous.

[0022] Furthermore, the step S2 includes the following steps: 1) firstly creating an attraction field for the end point and a repulsion field for the obstacle; 2) randomly selecting a position: randomly selecting a position in the sampling space as a potential sampling point q new ; 3) Calculate the potential field value: For randomly generated points q new , calculate its gravitational potential and repulsive potential. This usually involves the location information of the target point and the obstacle; 4) Calculate the total potential field value: If the potential field value meets the predetermined conditions, then accept q new As an effective random sampling point.

[0023] This is because by creating a gravitational field for the destination and a repulsive field for obstacles, random sampling points are more likely to be located toward the destination and away from obstacles, thereby improving the effectiveness and success rate of path planning and strengthening the algorithm's guidance. At the same time, the repulsive field ensures that random sampling points do not get too close to obstacles, reducing potential collision risks during path planning. Furthermore, the gravitational field helps guide the search process toward the destination, accelerating the algorithm's convergence and making path planning more rapid.

[0024] Furthermore, the step S3 includes the following steps: 1) q new Point construction cylinder: at random sampling points q new A cylinder is built at the position of q new , radius is 0.05, height is 0.2; 2) Perform the following detection on each obstacle: calculate the minimum distance between the obstacle and the cylinder; if the minimum distance ≤ 0.1, record a collision; 3) Process the collision result: if a collision is detected, return to step S2 and regenerate a random sampling point q new , and re-establish the cylinder for collision detection; if no collision is detected, the sampling point q new Valid, continue to execute subsequent path planning steps.

[0025] This is because by building a cylinder around the random sampling points and performing collision detection to avoid collisions, we can ensure that the sampling points are not located inside or near obstacles, thus avoiding collisions with obstacles during the path planning process. Effectively eliminating sampling points that are too close to obstacles helps generate a smoother and more direct path, reducing path complexity and detours, and improving the overall path quality. The principle is to calculate whether the horizontal and vertical distances of two cylinders overlap for collision detection.

[0026] Furthermore, the step S4 includes the following steps: 1) traversing the path tree: traversing all nodes in the current path tree; 2) calculating the distance: for each node, calculating the distance between it and the random sampling point q new 3) Compare and select the minimum distance: Compare all calculated distances and find the node with the minimum distance, which is the nearest node. q n ; 4) Update the tree structure: In the path tree, q n Marked as q new The parent node ofq n Add to the list of child nodes q new ;5)Store connection information: record q n and q new The connection information between them is used to facilitate subsequent path search and optimization.

[0027] This is because this step is crucial in the RRT algorithm for expanding the path tree and maintaining its structural integrity. This allows the algorithm to gradually explore the search space until it finds a feasible path from the starting point to the goal, ensuring its efficiency and robustness, making it applicable to a variety of complex path planning problems.

[0028] Furthermore, the step S5 includes the following steps: 1) Calculating the distance from the root to q n The known optimal path cost and add it to the cost from q n arrive q new The new path cost is q new The initial total path cost; 2) traversal q new Neighboring nodes: Check all nodes with q new directly connected nodes (neighboring nodes), which means checking q n All child nodes of q n 3) For each neighboring node, calculate the path from the root to the neighboring node, and then through the neighboring node to q new 4) Compare path costs: Compare each neighboring node to q new The path cost is compared with the current minimum total path cost; if a path cost through a neighboring node is found to be lower, the update q new 5) Reconnect: If a lower-cost path is found, then q new From the original parent node q n Disconnect it and connect it to a new parent node (i.e., a neighboring node with a lower path cost).

[0029] This is because this step continuously searches for and updates optimal paths, ultimately resulting in a path that is both longer and more efficient, improving path quality. It also reduces computational effort by promptly updating the parent node of a node, eliminating unnecessary path extensions in subsequent searches and calculations, improving the algorithm's operational efficiency. Furthermore, it enhances the algorithm's adaptability. By dynamically adjusting and optimizing strategies, the algorithm can adapt to complex planning environments and find effective paths even in densely populated obstacles or confined spaces.

[0030] Furthermore, the step S6 includes the following steps: 1) calculating q new and Q goal 2) Compare the calculated distance with the step size: if the distance is less than the step size, it is considered q new close enough Q goal , you can continue to step S7. If the distance is greater than or equal to the step size, it is considered that you need to get closer Q goal Therefore, the process returns to step S2 to perform new sampling.

[0031] This distance-based decision is a common practice and simplifies the implementation of the algorithm because it provides a clear criterion for deciding when to stop sampling and start connecting the final paths.

[0032] Furthermore, the step S7 includes the following steps: 1) collecting path points: collecting the path points from the starting point to the current point. q new All path points that make up the original unsmoothed path; 2) Applying the Gaussian smoothing algorithm: Gaussian smoothing is a common signal processing technique that uses Gaussian functions as weights to perform a weighted average of data points to reduce noise and fluctuations. During path smoothing, the Gaussian smoothing algorithm is applied to the original path points, calculating the weighted average of each point. This process typically involves the following steps:

[0033] a. Determine the parameters of the Gaussian function, such as the standard deviation (σ). The standard deviation determines the degree of smoothing. A larger σ will produce a smoother path, but may also lose path details.

[0034] b. For each point on the path, calculate the weighted average of its surrounding points using Gaussian weights. The weight decreases with distance; the farther away, the lower the weight.

[0035] c. Replace each original path point with its weighted average, or add new points to form the smoothed path.

[0036] 3) Generate a smoothed path: Generate a set of new smoothed path points by applying Gaussian smoothing to each original path point. These new smoothed path points should be smoother than the original path while still maintaining the overall shape and direction of the path. 4) Add to the smoothed path set: Add the processed smoothed path points to the new smoothed path set.

[0037] The principles and purpose of the above steps are as follows: To ensure the continuity and smoothness of the planned path, the Gaussian smoothing algorithm is used to eliminate the inflection points in the distal motion trajectory of the fracture, thereby achieving a smoother and more natural motion trajectory. By integrating the Gaussian smoothing algorithm, dynamic transitions can be optimized and potential mechanical stress can be reduced, so that the robot can avoid causing secondary injuries as much as possible during the resetting movement.

[0038] Furthermore, step S8 includes the following steps: 1) Integrate smoothed path points: Integrate the set of smoothed path points generated in step S7 into a continuous path. This may involve connecting adjacent smoothed points to ensure path continuity and smoothness. 2) Generate final path: Define the optimized smoothed path as the final path. This path should be the best feasible path from the starting point to the destination point, taking into account all constraints and planning objectives. 3) Output path: Output the final path in an appropriate data structure or format so that it can be used for resetting the task. 4) End path planning: After successfully generating the final path, the path planning algorithm terminates. At this point, the algorithm can release resources or prepare to initialize for the next path planning task.

[0039] This makes routine operations more convenient to implement.

[0040] In this method, the RRT* path planning algorithm is modified by modifying the extended node step of the RRT* code. The artificial potential field algorithm code is then written and properly adjusted to purposefully generate nodes toward the endpoint under the influence of the potential field. During this process, cylinders are constructed at the nodes to determine whether they collide with cylinders on the path. Finally, a smoothing algorithm is written to the end of the source code to obtain the final smooth path. Therefore, this algorithm integrates the artificial potential field algorithm into the RRT* algorithm to enhance the purposefulness of random point generation. It also incorporates a cylindrical envelope obstacle avoidance strategy and simulates the physiological conditions of bone fractures to enhance the algorithm's obstacle avoidance performance. Finally, a Gaussian smoothing algorithm is incorporated to eliminate inflection points in the distal fracture trajectory, resulting in a smoother and safer trajectory, providing surgeons with more accurate path guidance.

[0041] Furthermore, the connecting arm is provided with a tension and pressure sensor, which is in communication with the control center, so as to detect the tensile stress data of the connecting arm, determine its stress condition, and assist in realizing the telescopic control of the connecting arm.

[0042] Furthermore, the telescopic device is an electric cylinder device. In this way, the electric cylinder device realizes telescopic control by a motor driving a lead screw nut, making the control of the telescopic process and telescopic size more precise and stable. Moreover, the electric cylinder device itself is a mature existing technology, easy to implement and manufacture, and easy to operate accurately.

[0043] Furthermore, the two ends of the connecting arm are hinged by a Hooke's hinge and a fixing ring, which makes the structure more reliable and does not affect the motion control of the connecting arm in all directions.

[0044] In the robot of the present invention, a new algorithm strategy is adopted in the path planning module for path planning design, which can provide doctors with a better solution to avoid collisions when performing fracture reduction operations. Compared with other algorithms, the improved RRT* algorithm fused with the APF algorithm in this application has high purposefulness, reduces the generation of redundant points, and reduces the running time. The time required for clinical fracture reduction path planning is greatly reduced. The cylindrical envelope method is introduced, and then combined with the artificial repulsive field to generate a collision-free path. The Gaussian smoothing algorithm is used to eliminate the inflection points in the motion trajectory, resulting in a smoother and safer motion trajectory.

[0045] In summary, the present application solution can assist doctors in their operations, so as to better avoid accidental injuries through obstacle avoidance strategies, and better improve the reliability of the robot in the fracture rehabilitation assistance process. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a schematic structural diagram of the intelligently controlled six-link fracture reduction robot of the present invention.

[0047] Figure 2 This is a structural diagram of the control part of the intelligently controlled six-link fracture reduction robot of the present invention.

[0048] Figure 3 In order to verify the feasibility of the path planning algorithm used in this application, a planning path diagram obtained by using the existing RRT algorithm is shown.

[0049] Figure 4 In order to verify the feasibility of the path planning algorithm used in this application, a planning path diagram obtained by using the existing RRT* algorithm is used.

[0050] Figure 5 In order to verify the feasibility of the path planning algorithm used in this application, a planning path diagram was obtained using the existing A* algorithm.

[0051] Figure 6 In order to verify the feasibility of the path planning algorithm used in this application, a planning path schematic diagram was obtained using the improved RRT* algorithm in this application.

[0052] Figure 7 In order to verify the feasibility of the path planning algorithm used in this application, a planning time comparison chart of four algorithms is used.

[0053] Figure 8 In order to verify the feasibility of the path planning algorithm used in this application, a comparison diagram of path nodes of four algorithms is used.

[0054] Figure 9 In order to verify the feasibility of the path planning algorithm used in this application, four algorithms were used to generate node comparison diagrams.

[0055] Figure 10 In order to verify the feasibility of the path planning algorithm used in this application, a comparison chart of the path lengths of four algorithms is used.

[0056] Figure 11 for Figure 7-10 Table of average values ​​of the obtained data. DETAILED DESCRIPTION

[0057] The following is combined with the accompanying drawings and specific implementation methods.

[0058] A preferred embodiment is an intelligently controlled six-link fracture reduction robot, comprising two fixing rings 1 for mounting Kirschner wires, six connecting arms 2 that are evenly distributed along the circumference and hingedly arranged between the two fixing rings 1, any two adjacent connecting arms 2 being close together at one end and separated at the other end in an eight-shaped pattern, each connecting arm 2 being provided with a telescopic device 3, and further comprising a control center (PC end) 4, the control center 4 being communicatively connected to the telescopic device 3 and realizing telescopic control, the control center 4 being provided with a control module 5 for controlling the relative position path of the two fixing rings and a path planning module 6 for designing the control path, the path planning module 6 having an improved RRT* path planning algorithm; the improved RRT* path planning algorithm comprising the following steps:

[0059] Step S1: Initialize the map and set the force field value and other related parameters;

[0060] Step S2: Introduce the artificial potential field method to generate random sampling points q new ;

[0061] Step S3: Using the cylindrical envelope method, q new Build a cylinder and check whether a collision occurs; if a collision occurs, return to step S2 and reselect the sample; if no collision occurs, continue to step S4;

[0062] Step S4: Find the tree node closest to the random sampling point qn , and use this node as q new The parent node of

[0063] Step S5: Find q new The minimum total path and reconnect;

[0064] Step S6: Determination q new arrive Q goal Is the distance less than the step size? If not, return to step 2 for sampling; if so, continue to step S7;

[0065] Step S7: Using Gaussian smoothing algorithm for fitting processing, adding the processed points to the new smoothed path set;

[0066] Step S8: Generate the final smooth planning path and the path planning task ends.

[0067] Thus, this solution utilizes an improved path planning algorithm within the path planning module of a conventional intelligently controlled six-link fracture reduction robot. The proposed improved algorithm performs path planning on the PC side, and the generated path node pose information is then imported into the control module, which then controls the motor rotation to achieve controlled reduction.

[0068] During implementation, step S1 includes the following steps: 1) initializing the map, setting the map size to 2×2×2 cm, setting the starting point to (-1,-1,-1), and setting the end point to (1,1,1); 2) setting the radius and height of the cylindrical obstacle to 0.05 and 0.2, respectively, and the position information to (0, 0, -0.1), (-0.6, -0.6, 0.4), (0.6, -0.6, 0.4), (-0.6, 0.6, 0.4), (0.6, 0.6, 0.4), (-0.6, -0.6, -0.4), (0.6, -0.6, -0.4) (-0.6, 0.6, -0.4) (0.6, 0.6, -0.4). 3) Set the parameters of the artificial force field: the relative strength of attraction is 1, the relative strength of repulsion is 2, the repulsive influence range is 0.3, and the step size is 0.04.

[0069] When implemented, the step S2 includes the following steps: 1) first creating an attraction field for the end point and a repulsion field for the obstacle; 2) randomly selecting a position: randomly selecting a position in the sampling space as a potential sampling point q new ; 3) Calculate the potential field value: For randomly generated points q new, calculate its gravitational potential and repulsive potential. This usually involves the location information of the target point and the obstacle; 4) Calculate the total potential field value: If the potential field value meets the predetermined conditions, then accept q new As an effective random sampling point.

[0070] The step S3 comprises the following steps: 1) q new Point construction cylinder: at random sampling points q new A cylinder is built at the position of q new , with a radius of 0.05 and a height of 0.2; 2) Perform the following checks on each obstacle: Calculate the minimum distance between the obstacle and the cylinder. If the minimum distance is ≤ 0.1, record a collision; 3) Process the collision result: If a collision is detected, return to step S2 and regenerate a random sampling point q new , and re-establish the cylinder for collision detection. If no collision is detected, the sampling point q new Valid, continue to execute subsequent path planning steps.

[0071] The step S4 includes the following steps: 1) traversing the path tree: traversing all nodes in the current path tree. 2) calculating the distance: for each node, calculating the distance between it and the random sampling point. q new 3) Compare and select the minimum distance: Compare all calculated distances and find the node with the minimum distance, which is the nearest node. q n ; 4) Update the tree structure: In the path tree, q n Marked as q new The parent node of q n Add to the list of child nodes q new 5) Store connection information: record q n and q new The connection information between them is used to facilitate subsequent path search and optimization.

[0072] The step S5 includes the following steps: 1) Calculating the distance from the root to q n The known optimal path cost and add it to the cost from q n arriveq new The new path cost is q new The initial total path cost. 2) Traverse q new Neighboring nodes: Check all nodes with q new directly connected nodes (neighboring nodes), which means checking q n All child nodes of q n 3) For each neighboring node, calculate the distance from the root to the neighboring node, and then through the neighboring node to q new 4) Compare path costs: Compare each neighboring node to q new The path cost of is compared with the current minimum total path cost. If a path cost through a neighboring node is found to be lower, the q new 5) Reconnect: If a lower cost path is found, then q new From the original parent node q n Disconnect it and connect it to a new parent node (i.e., a neighboring node with a lower path cost).

[0073] The step S6 comprises the following steps: 1) calculating q new and Q goal 2) Compare the calculated distance with the step size: If the distance is less than the step size, it is considered q new close enough Q goal , you can continue to step S7. If the distance is greater than or equal to the step size, it is considered that you need to get closer Q goal Therefore, the process returns to step S2 to perform new sampling.

[0074] The step S7 includes the following steps: 1) collecting path points: collecting path points from the starting point to the current point q new All path points in the original unsmoothed path are considered. 2) Applying Gaussian smoothing: Gaussian smoothing is a common signal processing technique that uses Gaussian functions as weights to perform a weighted average of data points to reduce noise and fluctuations. During path smoothing, the Gaussian smoothing algorithm is applied to the original path points, calculating a weighted average for each point. This process typically involves the following steps:

[0075] a. Determine the parameters of the Gaussian function, such as the standard deviation (σ). The standard deviation determines the degree of smoothing. A larger σ will produce a smoother path, but may also lose path details.

[0076] b. For each point on the path, calculate the weighted average of its surrounding points using Gaussian weights. The weight decreases with distance; the farther away, the lower the weight.

[0077] c. Replace each original path point with its weighted average, or add new points to form the smoothed path.

[0078] 3) Generate a smoothed path: Generate a set of new smoothed path points by applying Gaussian smoothing to each original path point. These new smoothed path points should be smoother than the original path while still maintaining the overall shape and direction of the path. 4) Add to the smoothed path set: Add the processed smoothed path points to the new smoothed path set.

[0079] Step S8 includes the following steps: 1) Integrate smoothed path points: Integrate the set of smoothed path points generated in step S7 into a continuous path. This may involve connecting adjacent smoothed points to ensure path continuity and smoothness. 2) Generate final path: Define the optimized smoothed path as the final path. This path should be the best feasible path from the starting point to the destination point, taking into account all constraints and planning objectives. 3) Output path: Output the final path in an appropriate data structure or format so that it can be used for resetting the task. 4) End path planning: After successfully generating the final path, the path planning algorithm terminates. At this point, the algorithm can release resources or prepare to initialize for the next path planning task.

[0080] In this method, the RRT* path planning algorithm is modified by modifying the extended node step of the RRT* code. The artificial potential field algorithm code is then written and properly adjusted to purposefully generate nodes toward the endpoint under the influence of the potential field. During this process, cylinders are constructed at the nodes to determine whether they collide with cylinders on the path. Finally, a smoothing algorithm is written to the end of the source code to obtain the final smooth path. Therefore, this algorithm integrates the artificial potential field algorithm into the RRT* algorithm to enhance the purposefulness of random point generation. It also incorporates a cylindrical envelope obstacle avoidance strategy and simulates the physiological conditions of bone fractures to enhance the algorithm's obstacle avoidance performance. Finally, a Gaussian smoothing algorithm is incorporated to eliminate inflection points in the distal fracture trajectory, resulting in a smoother and safer trajectory, providing surgeons with more accurate path guidance.

[0081] The connecting arm is also provided with a tension and pressure sensor 7, which is connected to the control center for communication. In this way, the tensile stress data of the connecting arm can be detected to determine its stress condition and assist in realizing the telescopic control of the connecting arm.

[0082] The telescopic device 3 is an electric cylinder device. Such an electric cylinder device realizes telescopic control by a motor driving a lead screw nut, making the control of the telescopic process and telescopic size more precise and stable. Moreover, the electric cylinder device itself is a mature existing technology, easy to implement and manufacture, and easy to operate accurately.

[0083] The two ends of the connecting arm 2 are hinged by a Hooke's hinge 8 and a fixing ring, which makes the structure more reliable and does not affect the motion control of the connecting arm in all directions.

[0084] In the robot of the present invention, a new algorithm strategy is adopted in the path planning module for path planning design, which can provide doctors with a better solution to avoid collisions when performing fracture reduction operations. Compared with other algorithms, the improved RRT* algorithm fused with the APF algorithm in this application has high purposefulness, reduces the generation of redundant points, and reduces the running time. The time required for clinical fracture reduction path planning is greatly reduced. The cylindrical envelope method is introduced, and then combined with the artificial repulsive field to generate a collision-free path. The Gaussian smoothing algorithm is used to eliminate the inflection points in the motion trajectory, resulting in a smoother and safer motion trajectory.

[0085] To verify the feasibility of the path planning algorithm proposed in this application, a verification experiment was designed and compared with the other three commonly used path planning algorithms (RRT, RRT*, A*). Under ideal conditions, in order to simplify the algorithm, the distal end of the fracture is regarded as a mass point during the simulation process, but the cylindrical bounding box method is used in the algorithm to simulate its spatial position relationship. The map environment size is set to 2cm×2cm×2cm, the initial position of the center of the distal fracture section is (-1,-1,-1)cm, and the target position is (0.9,0.9,0.9)cm, that is, the center position of the proximal fracture section. The complex environment is set to 9 cylinders with a radius of 0.05cm and a height of 0.2cm to ensure the single variable principle of the experiment.

[0086] In this complex environment, multiple tests were conducted on RRT, RRT*, A* and the improved algorithm of this application to verify the stability and security of the improved algorithm of this application. The paths planned by different algorithms are as follows: Figure 3-6 shown.

[0087] In the comparative test, the path planning time, the number of path nodes, the number of generated nodes and the path length are used as evaluation indicators. The results are as follows: Figure 7-10 As shown in the figure, due to the strong randomness of the RRT* algorithm and the RRT algorithm during sampling, the path planning time, generated nodes, and path length vary widely, with poor stability and many inflection points. However, the improved RRT* algorithm proposed in this application reduces redundant nodes in the path, resulting in a smaller range of variation in path planning time and path length, good stability, and a smoother path. Experimental results show that the improved RRT* algorithm can obtain a feasible smooth path with a smaller time cost and shorter path length, and can safely reach the target location. It can effectively improve path planning efficiency and meet the needs of fracture reduction.

[0088] Figure 11 The table shows Figure 7-10 The average of the items shown. Data analysis shows that the reset path planning time used by the improved RRT is significantly shortened by 99.3% compared to A* and 94.3% compared to RRT. In addition, the number of randomly generated points of the improved RRT is significantly reduced by 98.7% compared to A* and 96.8% compared to RRT. In terms of path length, the path length of the enhanced RRT is comparable to that of A*, only slightly longer than RRT, but significantly better than RRT. This indicates that its path efficiency is at a relatively high level. Regarding the number of path points generated, the number of path points of the improved RRT is lower than that of RRT, but higher than that of RRT and A*. This shows that the improved RRT* achieves better optimization while ensuring the smoothness and accuracy of the path.

Claims

1. An intelligently controlled six-link fracture reduction robot, comprising two fixed rings for mounting Kirschner wires, six connecting arms hingedly disposed between the two fixed rings and uniformly distributed along the circumference, wherein any two adjacent connecting arms are brought together at one end and separated at the other end to form an "eight" shape, each connecting arm being provided with a telescopic device, and further comprising a control center, the control center being communicatively connected to the telescopic device and implementing telescopic control, the control center being provided with a control module for controlling the relative position path of the two fixed rings and a path planning module for designing the control path, characterized in that: The path planning module has an improved RRT* path planning algorithm; the improved RRT* path planning algorithm includes the following steps: Step S1: Initialize the map and set the force field numerical parameters; Step S2: Introduce the artificial potential field method to generate random sampling points q new ; Step S3: Using the cylindrical envelope method, q new Build a cylinder and check whether a collision occurs; if a collision occurs, return to step S2 and reselect the sample; if no collision occurs, continue to step S4; Step S4: Find the tree node closest to the random sampling point q n , and use this node as q new The parent node of Step S5: Find q new The minimum total path and reconnect; Step S6: Determination q new arrive Q goal Is the distance less than the step size? If not, return to step 2 for sampling; if so, continue to step S7; Step S7: Using Gaussian smoothing algorithm for fitting processing, adding the processed points to the new smoothed path set; Step S8: Generate the final smooth planning path and the path planning task ends; The step S1 includes the following sub-steps: 1) initializing the map, setting the map size to 2×2×2 cm, setting the starting point to (-1,-1,-1), and setting the end point to (1,1,1); 2) setting the radius and height of the cylindrical obstacle to 0.05 and 0.2 respectively, and the position information to (0, 0, -0.1), (-0.6, -0.6, 0.4), (0.6, -0.6, 0.4), (-0.6, 0.6, 0.4), (0.6, 0.6, 0.4), (-0.6, -0.6, -0.4), (0.6, -0.6, -0.4) (0.6, 0.6, -0.4) -0.4); 3) Set the parameters of the artificial force field: the relative strength of attraction is 1, the relative strength of repulsion is 2, the repulsive influence range is 0.3, and the step size is 0.04; The step S2 includes the following steps: 1) firstly creating an attraction field for the end point and a repulsion field for the obstacle; 2) randomly selecting a position in the sampling space as a potential sampling point q new ; 3) Calculate the potential field value for the randomly generated point q new , calculate its attractive potential and repulsive potential; 4) calculate the total potential field value; if the potential field value meets the predetermined conditions, then accept q new As an effective random sampling point; The step S3 comprises the following steps: 1) q new Point to build a cylinder, that is, at random sampling points q new A cylinder is built at the position of q new , radius is 0.05, height is 0.2; 2) Perform the following detection on each obstacle and calculate the minimum distance between the obstacle and the cylinder; if the minimum distance is ≤ 0.1, record a collision; 3) Process the collision result. If a collision is detected, return to step S2 and regenerate a random sampling point q new , and re-establish the cylinder for collision detection; if no collision is detected, the sampling point q new If it is valid, continue to execute the subsequent path planning steps; The step S4 includes the following steps: 1) traversing the path tree, traversing all nodes in the current path tree; 2) calculating the distance, for each node, calculating its distance to the random sampling point q new 3) Compare and select the minimum distance, that is, compare all calculated distances and find the node corresponding to the minimum distance, which is the nearest node q n ; 4) Update the tree structure. In the path tree, q n Marked as q new The parent node of q n Add to the list of child nodes q new ; 5) Store connection information and record q n and q new The connection information between them is used to facilitate subsequent path search and optimization; The step S5 includes the following steps: 1) Calculating the distance from the root to q n The known optimal path cost and add it to the cost from q n arrive q new The new path cost is q new The initial total path cost; 2) traversal q new Check all neighboring nodes of q new Directly connected nodes, i.e. check q n All child nodes of q n 3) For each neighboring node, calculate the path from the root to the neighboring node, and then through the neighboring node to q new 4) Compare the path costs and compare each neighboring node to q new The path cost is compared with the current minimum total path cost; if a path cost through a neighboring node is found to be lower, the q new The minimum total path cost is calculated and the new path is recorded; 5) Reconnect. If a lower cost path is found, the q new From the original parent node q n Disconnect it and connect it to the new parent node; The step S6 comprises the following steps: 1) calculating q new and Q goal 2) Compare the calculated distance with the step size: if the distance is less than the step size, it is considered q new close enough Q goal , then continue to step S7; if the distance is greater than or equal to the step length, it is considered that it is necessary to get closer Q goal The sampling point is thus returned to step S2 for new sampling; The step S7 includes the following steps: 1) collecting path points: collecting path points from the starting point to the current point q new All path points of , which constitute the original non-smooth path; 2) Apply the Gaussian smoothing algorithm to calculate the weighted average of each point, which includes the following steps: a. Determine the parameters of the Gaussian function, including the standard deviation (σ); b. For each point on the path, calculate the weighted average of its surrounding points using the Gaussian weight; c. Replace each original path point with its weighted average, or add new points to form the smoothed path; 3) Generate a smooth path by applying Gaussian smoothing to each original path point to generate a new set of smoothed path points; 4) Adding the smooth path set, that is, adding the processed smooth path points to the new smooth path set; The step S8 includes the following steps: 1) integrating smooth path points: integrating the set of smooth path points generated in step S7 into a continuous path; 2) generating a final path: defining the optimized smooth path as the final path, which is the best feasible path from the starting point to the target point; 3) outputting the path, that is, outputting the final path in an appropriate data structure or format so that it can be used for the reset task; 4) ending the path planning task, that is, after successfully generating the final path, the task of the path planning algorithm ends.

2. The intelligent controlled six-link fracture reduction robot according to claim 1, characterized in that: The connecting arm is also provided with a tension and pressure sensor, which is communicatively connected to the control center.

3. The intelligent controlled six-link fracture reduction robot according to claim 1, characterized in that: The telescopic device is an electric cylinder device.

4. The intelligent controlled six-link fracture reduction robot according to claim 1, characterized in that: The two ends of the connecting arm are hinged by Hooke hinge and fixed ring.

Images