CN-122005081-A - Concentric tube robot path planning method and system for laryngopharynx minimally invasive surgery

CN122005081ACN 122005081 ACN122005081 ACN 122005081ACN-122005081-A

Abstract

The invention provides a concentric tube robot path planning method and system for a laryngopharynx minimally invasive surgery. The invention introduces directional distance field and robot skeleton information, wherein scalar value of the directional distance field is composed of minimum distance from point to surface and inner and outer positions of the directional distance field, the directional distance field is used for describing shortest signed distance from any point in space to surrounding tissue surface, and reflecting obstacle position and safety margin, and the robot skeleton is obtained by discretizing sampling of a laryngopharynx concentric tube robot along the arc length direction of a sleeve under given joint configuration and used for guiding path morphology. The multi-strategy tree expansion is used for selecting three tree expansion strategies of standard expansion, gradient guiding expansion and intelligent sample expansion according to the set probability so as to achieve the requirements of sampling diversity and path constraint, thereby improving the accessibility and continuity of path generation and adapting to the limited environment of natural cavity tract operation.

Inventors

LIN ZHENGYING
CHEN JINGSHANG
LIANG YONGHUI

Assignees

福州大学
福建中医药大学附属第三人民医院

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (10)

1. The concentric tube robot path planning method for the laryngopharynx minimally invasive surgery is characterized by comprising the following steps of: Step S1, initializing a path planning environment, and converting a three-dimensional grid model of a laryngopharynx operation scene into a directed distance field; step S2, setting initial joint Target joint Joint space boundaries And calculating a skeleton point set corresponding to the target joint ; Step S3, generating random joints by proposing a multi-mode self-adaptive sampling strategy Performing gradient guidance expansion, intelligent sample expansion or standard expansion according to probability selection, wherein the standard expansion and the gradient guidance expansion adopt a self-adaptive target bias sampling strategy, and the intelligent sample expansion adopts an intelligent sample sampling strategy; S4, executing multi-strategy tree expansion to generate a new joint, selecting an expansion mode according to a sampling point strategy, and searching distances in a random tree Nearest neighbor node And slave according to the selected extension policy Orientation of Generating new joints ; S5, performing skeleton similarity constraint check on the new joint, calculating the similarity distance between the new joint skeleton and the target skeleton, and judging whether the dynamic threshold constraint and the collision detection constraint are met; step S6, optimizing a random tree structure by adopting a self-adaptive rerouting strategy, and dynamically adjusting the rerouting execution frequency and the searching radius according to the planning state; And S7, after the feasible path is found, performing multi-level path optimization, including global rerouting optimization, path smoothing, local optimization and path simplification.
2. The concentric tube robotic path planning method for use in minimally invasive laryngopharynx surgery as claimed in claim 1, wherein step S1 includes the following: in step S1, voxel processing is carried out on a three-dimensional STL grid model of the laryngopharyngeal operation environment, a regular grid is established in a three-dimensional space, and the shortest directional distance from each grid point to the surface of the obstacle is calculated Wherein the points in free space satisfy The points inside the obstacle are satisfied The points of the obstacle surface are satisfied 。
3. The concentric tube robotic path planning method for use in minimally invasive laryngopharynx surgery as claimed in claim 2, wherein step S2 further includes the following: in step S2, according to the kinematic model of the laryngopharynx concentric tube robot, the target joint is aligned along the arc length direction of the sleeve Performing discretization sampling, calculating three-dimensional coordinates of each target point in Cartesian space to form a target skeleton point set Wherein The number of sampling points for the skeleton.
4. The concentric tube robotic path planning method for use in minimally invasive laryngopharynx surgery as claimed in claim 1, wherein step S3 includes the following: Step S31, gradient guidance expansion, intelligent sample expansion or standard expansion is performed according to probability selection: When no path is found, standard extension probability is performed Gradient guided extension probability Intelligent sample expansion is not performed; when the path is found, standard extension probability is executed Gradient guided extension probability Performing intelligent sample extension probabilities ; Step S32, providing a multi-mode self-adaptive sampling strategy in a sampling stage, and improving the sampling quality through fusion of standard sampling, gradient guiding sampling, intelligent sample sampling and three modes, wherein the corresponding sampling modes are carried out according to an expansion strategy: adaptive target bias sampling strategy for standard expansion and gradient guided expansion, and target bias probability thereof Dynamically adjusting according to the path searching progress: ; Wherein, the For the current number of iterations, The maximum iteration number; Sampling point The generation strategy of (1) is as follows: ; Wherein the method comprises the steps of In order to be a uniform random number, In order to achieve a target joint, Is joint space Uniformly and randomly sampling the inner part; step S33, intelligent sample sampling strategy is used for intelligent sample expansion, when the feasible path is found, the current optimal path is selected Generating an intelligent sampling point set The method comprises the steps of including a neighborhood sampling point and a path midpoint sampling point; For any node on the path Neighborhood sampling point thereof The calculation formula is as follows: ; wherein the random offset vector Each dimension of (a) is independently from the interval And (3) inner uniform sampling: ; Wherein, the As the dimension of the joint, Sampling radius for the neighborhood; Step S34 for optimal Path Any pair of adjacent path points in the set of the sampling points are generated, and the set of the sampling points are used for improving the continuity of sampling distribution and the smoothness of the path The calculation mode of (a) is as follows: ; Wherein the method comprises the steps of Configuring for the ith joint in the path; configuring the (i+1) th joint in the path; Step S35 Intelligent sampling set From neighborhood sampling points And midpoint sampling point And randomly selecting sampling points from the set when intelligent sample expansion is carried out: ; Wherein the method comprises the steps of 。
5. The concentric tube robotic path planning method for use in minimally invasive laryngopharynx surgery as claimed in claim 1, wherein step S4 includes the following: step S41, a new node generating mode by tree expansion is as follows: the standard expansion strategy adopts a linear propulsion mode of step length constraint to generate a new node, and the calculation formula is as follows: ; Wherein the method comprises the steps of Is a step size parameter; step S42, calculating an obstacle avoidance vector by using gradient information of a directed distance field by a gradient guidance strategy to realize obstacle avoidance expansion based on environment geometry, wherein a new node calculation mode is as follows: ; Wherein the method comprises the steps of For the weight coefficients towards the random point direction, Is a weight coefficient of the obstacle avoidance vector; representing a unit vector in the direction of the random point, Representing an obstacle avoidance vector; step S43 from the current node Pointing to random nodes Direction vector of (a) The calculation formula is as follows: ; Candidate random joint Corresponding skeleton point Directed distance field value of (2) If it meets Calculating obstacle avoidance vector, otherwise, letting Wherein the gradient is calculated by adopting a center difference method to obtain: ; Wherein the method comprises the steps of , , Is calculated by the same way ; ; Step S44, normalizing the gradient to obtain the obstacle avoidance direction: ; Step S44, the obstacle avoidance intensity coefficient is as follows: ; The obstacle avoidance vectors are mapped to joint space by jacobian: ; Wherein the method comprises the steps of As a pseudo-inverse of the jacobian matrix, ; Wherein the intelligent sample expansion strategy is consistent with the standard expansion mode, but random sampling points are adopted From the intelligent sample set.
6. The concentric tube robotic path planning method for use in minimally invasive laryngopharynx surgery as claimed in claim 1, wherein step S5 includes the following: step S51, representing the robot shape as a point sequence consisting of a plurality of discrete skeleton points: ; ; Wherein the method comprises the steps of -A set of skeletal points of the target shape; -a set of skeleton points of the shape corresponding to the new joint; The skeleton similarity distance is defined as: ; Wherein the method comprises the steps of ; The dynamic threshold is defined as: Wherein the method comprises the steps of Is a constant coefficient; At the same time satisfy , The new node is considered to satisfy the shape constraint and collision detection constraint and is added to the random tree, otherwise it is discarded.
7. The concentric tube robotic path planning method for use in minimally invasive laryngopharynx surgery as claimed in claim 1, wherein step S6 includes the following: step S61, the adaptive rerouting strategy includes: Immediately performing rerouting upon finding a new path; performing rewiring once every 50 iterations when no path is found; based on adaptive frequency parameters when a path has been found Performing rewiring; wherein the rerouting search radius is defined as: ; Wherein the method comprises the steps of In order to be the number of nodes, Is the joint space dimension; Wherein the method comprises the steps of As a theoretical constant, defined as: ; Wherein the method comprises the steps of Is joint space Is defined by the volume of (2); for the limit of the search radius, it is defined as: ; Wherein the method comprises the steps of Is a constant coefficient.
8. The concentric tube robotic path planning method for use in minimally invasive laryngopharynx surgery as claimed in claim 1, wherein step S7 includes the following: in step S7, the multi-level path optimization includes: Global rerouting optimization, namely searching an optimal father node for the path node in the whole tree again; path smoothing, namely attempting to skip intermediate nodes to directly connect non-adjacent nodes; Local optimization, namely local adjustment is carried out in the node neighborhood to reduce the path cost; wherein the path cost function is defined as: ; Wherein the first item Representing geometric distance, second term Third term as skeletal shape deviation cost Is a safety margin cost based on SDF.
9. A concentric tube robotic path planning system for a laryngopharynx minimally invasive procedure comprising an electronic device, wherein the electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements a concentric tube robotic path planning method for a laryngopharynx minimally invasive procedure as claimed in any one of claims 1 to 8 when executing the computer program.
10. A concentric tube robot path planning system for a laryngopharyngeal minimally invasive surgery comprising a computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements a concentric tube robot path planning method for a laryngopharyngeal minimally invasive surgery as claimed in any one of claims 1 to 8.

Description

Concentric tube robot path planning method and system for laryngopharynx minimally invasive surgery Technical Field The invention provides a concentric tube robot path planning method and system for laryngopharynx minimally invasive surgery, and relates to the technical field of medical robot path planning. Background In minimally invasive transoral access to the laryngopharynx, the medical robot needs to reach the target area along the natural orifice, and its path planning must meet the requirements of accessibility from the oral entrance to the laryngopharynx target lesion, the robot maintains sufficient safety margin throughout the orifice to avoid collisions, and the planned path matches the shape characteristics of the robot itself. The laryngopharynx concentric tube robot is formed by nesting a plurality of pre-bent elastic tube bodies, the tubes can be independently pushed and pulled axially and rotated circumferentially, and the change of the pose of the tail end (clamping scalpel) in space is realized by controlling the relative movement of the tubes, so that the laryngopharynx concentric tube robot is suitable for a narrow natural cavity environment and completes the corresponding laryngopharynx position operation. The traditional path planning method, such as grid searching methods of A, dijkstra and the like, can expand the searching space rapidly along with the increase of the joint freedom, so that the grid searching is difficult to realize real-time path planning in the joint space of the high-dimensional flexible robot. The methods such as RRT and RRT based on random sampling can process a high-dimensional space, but a single tree expansion strategy is generally adopted, so that environmental constraint and motion characteristics cannot be considered, and the generated paths are discontinuous in a local area or do not meet the requirement of executable. Disclosure of Invention In view of the above, in order to make up for the blank and the deficiency of the prior art, the invention provides a concentric tube robot path planning method and system for a laryngopharynx minimally invasive surgery. The invention introduces directional distance field (SDF) and robot skeleton (skeleton) information, wherein scalar values of the directional distance field are formed by the minimum distance from a point to a surface and the inner and outer positions of the directional distance field, are used for describing the shortest signed distance from any point in a space to the surface of surrounding tissues, reflect the position and safety margin of an obstacle, and are obtained by discretizing and sampling a laryngopharynx concentric tube robot along the arc length direction of a sleeve under the given joint configuration and are used for guiding the path form. The multi-strategy tree expansion is used for selecting three tree expansion strategies of standard expansion, gradient guiding expansion and intelligent sample expansion according to the set probability so as to achieve the requirements of sampling diversity and path constraint, thereby improving the accessibility and continuity of path generation and adapting to the limited environment of natural cavity tract operation. The invention provides a concentric tube robot path planning method and system for laryngopharynx minimally invasive surgery, comprising the following steps: The invention provides a concentric tube robot path planning method for a laryngopharynx minimally invasive surgery, which is characterized by comprising the following steps of: Step S1, initializing a path planning environment, and converting a three-dimensional grid model of a laryngopharynx operation scene into a directed distance field; step S2, setting initial joint Target jointJoint space boundariesAnd calculating a skeleton point set corresponding to the target joint; Step S3, generating random joints by proposing a multi-mode self-adaptive sampling strategyPerforming gradient guidance expansion, intelligent sample expansion or standard expansion according to probability selection, wherein the standard expansion and the gradient guidance expansion adopt a self-adaptive target bias sampling strategy, and the intelligent sample expansion adopts an intelligent sample sampling strategy; S4, executing multi-strategy tree expansion to generate a new joint, selecting an expansion mode according to a sampling point strategy, and searching distances in a random tree Nearest neighbor nodeAnd slave according to the selected extension policyOrientation ofGenerating new joints; S5, performing skeleton similarity constraint check on the new joint, calculating the similarity distance between the new joint skeleton and the target skeleton, and judging whether the dynamic threshold constraint and the collision detection constraint are met; step S6, optimizing a random tree structure by adopting a self-adaptive rerouting strategy, and dynamically adjusting the rerouting execution frequenc