Search

CN-121979271-A - Intelligent inspection robot path dynamic optimization method and system

CN121979271ACN 121979271 ACN121979271 ACN 121979271ACN-121979271-A

Abstract

The application relates to the technical field of artificial intelligence and robots, discloses a dynamic path optimization method and a dynamic path optimization system for an intelligent inspection robot, and aims to solve the problems of path interruption, low efficiency and resource unbalance caused by dependence on a static map, response delay, frequent re-planning and task semantic information neglecting in the prior art. The method involves the steps of acquiring environmental data in real time by using a multi-mode sensor, constructing a local occupation grid map, generating an initial path by combining a global topological map and an improved RRT (remote radio unit) algorithm, and calculating a path deviation index PII and an obstacle approximation rate OAR in real time. By the technology, low-delay and high-robustness path optimization under a dynamic environment is realized, the response speed is remarkably improved, the rescheduling times are reduced by 65%, the routing inspection path is shortened by 12.3%, the steering action is reduced by 21.5%, and the running stability and safety are ensured. The improved strategies are consistent with the research results of path planning and optimization of the intelligent inspection robot, and the method has broad prospects in engineering application.

Inventors

  • WU GAO
  • Duan Qiuhai

Assignees

  • 广东华能机电集团有限公司

Dates

Publication Date
20260505
Application Date
20260204

Claims (10)

  1. 1. The intelligent routing inspection robot path dynamic optimization method is characterized by comprising the following steps of: Acquiring environment sensing data of a patrol area in real time through a multi-mode sensor group, wherein the environment sensing data comprises space point cloud data, obstacle thermal characteristic information and visible light images; based on environment perception data, constructing a local occupation grid map by taking the current position of the robot as a center, and identifying a passable area, a temporary barrier and a potential dangerous area according to the difference between echo intensity and surface normal direction, so as to form an incremental environment model with semantic tags; combining a pre-stored global topological map and a task scheduling instruction to generate a reference navigation path from a current point to a next target point; comparing the incremental environmental model with the predicted section of the reference navigation path in real time, calculating a path deviation index and an obstacle approximation rate, and generating a path feasibility score based on weighted fusion of the path deviation index and the obstacle approximation rate; Triggering a path re-planning process when the path feasibility score is continuously lower than a preset threshold value; In the path re-planning process, a new optimized path is generated by taking the minimized path length, the accumulated change quantity of the steering angle and the degree of approaching to a high risk area as objective functions; And carrying out track smoothing on the new optimized path, and driving the robot to execute track tracking through a model predictive control algorithm.
  2. 2. The method for dynamically optimizing a path of an intelligent inspection robot according to claim 1, wherein the collecting environmental perception data of an inspection area in real time by a multi-mode sensor group comprises: synchronously acquiring original data by using a sensing array formed by a laser radar, a depth camera and an infrared sensor; And performing time stamp alignment and coordinate system one calibration on the original data, and outputting the structured three-dimensional environment state representation as the environment perception data.
  3. 3. The method for dynamically optimizing a path of an intelligent inspection robot according to claim 1, wherein calculating a path deviation index and an obstacle approximation rate comprises: projecting the current pose of the robot to the nearest path node on the reference navigation path, and calculating the ratio of the transverse deviation to the local perception window radius as the path deviation index; And estimating a projection component of the relative speed of the front dynamic obstacle in the tangential direction of the path by using Kalman filtering as the obstacle approximation rate.
  4. 4. The method for dynamically optimizing a path of an intelligent inspection robot according to claim 1, wherein triggering the path re-planning procedure when the path feasibility score is continuously below a preset threshold value comprises: Setting a two-stage trigger mechanism, and starting a background lightweight path pre-calculation process to generate a candidate path set when a primary early warning threshold is met; and when the emergency re-planning threshold is met, selecting an optimal path from the candidate path set to switch, or directly calling a path optimization engine to generate a brand new optimized path.
  5. 5. The method of claim 1, wherein generating a new optimized path with a minimized path length, a steering angle cumulative change, and a degree of approach to a high risk area as objective functions comprises: constructing a mixed integer quadratic programming problem model, wherein constraint conditions of the mixed integer quadratic programming problem model comprise the maximum linear speed, the maximum angular speed, the minimum turning radius and the safety interval between the mixed integer quadratic programming problem model and a dangerous area; And calling a solver to solve the mixed integer quadratic programming problem model, and outputting a path node sequence meeting all constraints as the new optimized path.
  6. 6. The method for dynamically optimizing a path of an intelligent inspection robot according to claim 1, wherein performing a trajectory smoothing process on the new optimized path comprises: B spline curve fitting is carried out on the path node sequence of the new optimized path, and a continuous and conductive reference track is generated; the control point spacing of the B spline is automatically adjusted according to the curvature of the path, smaller spacing is adopted in the high-curvature section to keep shape fidelity, and the spacing is increased in the straight line section to reduce data redundancy.
  7. 7. The method for dynamically optimizing a path of an intelligent inspection robot according to claim 1, further comprising, in a multi-robot collaboration scenario: receiving virtual repulsive force field information of an adjacent robot, and superposing the virtual repulsive force field information to a local occupied grid map of the adjacent robot; calculating the ratio of the space intersection length of the current path and the virtual repulsive force field to the path length of the current path and the virtual repulsive force field as a collision risk index; the collision risk index is incorporated into the calculation of the path feasibility score to resolve path collisions preferentially through a negotiation mechanism.
  8. 8. The intelligent patrol robot path dynamic optimization method according to claim 7, wherein incorporating a collision risk index into the calculation of path feasibility score comprises: adding a weighted term to the path feasibility score that is related to the collision risk index; And when the path feasibility score is lower than the preset threshold value and the conflict risk index is higher than a conflict judgment threshold value, starting a distributed negotiation protocol instead of directly triggering path re-planning.
  9. 9. An intelligent inspection robot path dynamic optimization system, comprising: The environment sensing module is used for collecting environment sensing data of the inspection area in real time through the multi-mode sensor group, wherein the environment sensing data comprises space point cloud data, obstacle thermal characteristic information and visible light images; the local topology modeling module is connected with the environment perception module and is used for constructing a local occupation grid map by taking the current position of the robot as a center based on the environment perception data, and identifying a passable area, a temporary obstacle and a potential dangerous area according to the difference between the echo intensity and the surface normal vector, so as to form an incremental environment model with semantic tags; the global path guiding module is used for combining a pre-stored global topological map and a task scheduling instruction to generate a reference navigation path from a current point to a next target point; The dynamic risk assessment module is connected with the local topology modeling module and the global path guiding module and is used for comparing the incremental environment model with the predicted section of the reference navigation path in real time, calculating a path deviation index and a barrier approximation rate, and generating a path feasibility score based on weighted fusion of the path deviation index and the barrier approximation rate; The re-planning decision module is connected with the dynamic risk assessment module and is used for triggering a path re-planning process when the path feasibility score is continuously lower than a preset threshold value; The path optimization engine is used for generating a new optimized path by taking the minimized path length, the steering angle accumulated change quantity and the degree of approaching to the high risk area as objective functions in the path re-planning process; and the track smoothing and tracking module is connected with the path optimization engine and is used for carrying out track smoothing processing on the new optimized path and driving the robot to execute track tracking through a model predictive control algorithm.
  10. 10. The intelligent patrol robot path dynamic optimization system according to claim 9, further comprising a communication coordination interface module for realizing high-speed data interaction between each functional module, wherein in a multi-robot collaboration scenario, the communication coordination interface module is further used for broadcasting local virtual repulsive force field information and receiving virtual repulsive force field information of a neighboring robot, and the dynamic risk assessment module is further used for calculating a ratio of a space intersection length of a current path and the virtual repulsive force field to a path length of the current path and the virtual repulsive force field as a collision risk index, and incorporating the collision risk index into calculation of the path feasibility score.

Description

Intelligent inspection robot path dynamic optimization method and system Technical Field The invention belongs to the technical field of artificial intelligence and robots, and particularly relates to a method and a system for dynamically optimizing an intelligent inspection robot path. Background With the continuous improvement of the intelligent manufacturing and industrial automation level, intelligent robots are increasingly widely applied in complex industrial environments. Particularly, in key scenes such as equipment inspection, safety monitoring and fault early warning, the intelligent robot plays a vital role. For example, the intelligent inspection robot can be provided with a high-precision sensor and a camera to perform autonomous inspection and detection in dangerous or difficult-to-enter environments, so that inspection efficiency and accuracy are remarkably improved. The intelligent inspection robot is used as an important component of high-end equipment, integrates various functional modules such as sensing, decision making and execution, can replace manual work to complete periodic inspection tasks in an unattended or high-risk environment, and remarkably improves the safety and efficiency of operation and maintenance. The path planning is used as one of the core capacities of autonomous movement of the robot, and directly relates to the integrity of inspection coverage, the timeliness of task execution and the economy of energy consumption. In the field of industrial inspection, a path optimization technology in a dynamic environment has become a key break for realizing efficient inspection. The technology aims at carrying out online adjustment on the travel route of the robot according to real-time perceived environmental information, task priority change and target area state fluctuation so as to adapt to continuously changing operation requirements. Under an ideal state, the dynamic optimization of the path needs to consider the overall accessibility and the local response capability, and the repeated walking or path oscillation is avoided while the full coverage is ensured, so that the multi-objective balance of time, energy consumption and detection quality is realized. The method has the advantages that the method is not easy to deal with the problem of path interruption caused by sudden obstacles or temporary forbidden areas due to the fact that a plurality of methods depend on a preset static map and a fixed inspection sequence, is slow in environment change response mechanism, lacks instant analysis and quick re-planning capability of sensor data streams, and causes robots to frequently stop or bypass inefficient paths under dynamic interference, the path updating strategy is easy to consider the shortest geometrical distance, task semantic information such as equipment heating state and failure risk level is ignored, key area missed inspection or resource allocation imbalance is caused, and the method is not easy to occur path conflict, area overlapping inspection or communication congestion phenomenon among robots under a multi-machine collaborative scene, so that the operation efficiency of the whole system is seriously affected. These problems are particularly prominent in application scenes with complex structures and extremely high safety requirements, such as large-scale transformer stations, petrochemical plant areas and the like, and the problems become technical bottlenecks for restricting the practical level improvement of the intelligent inspection system. Disclosure of Invention The invention aims to make up the defects of the prior art, provides a method and a system for dynamically optimizing the path of an intelligent inspection robot, and can effectively solve the problems in the background art. In the application fields of high-end equipment manipulators and robots, particularly in intelligent inspection tasks in complex industrial environments, a traditional path planning method generally depends on a static map and a preset route, and is difficult to adapt to actual conditions of frequent change of field device layout, dynamic occurrence of obstacles and concurrent execution of multiple tasks. The method generally generates an initial path based on static diagram search algorithms such as A or Dijkstra and the like, and lacks effective fusion capability on real-time environment perception data, so that the robot frequently triggers emergency stop or detours in the running process, and the inspection efficiency and the system reliability are seriously affected. In addition, the prior art fails to construct a quantitative association model between the environmental change degree and the path re-planning trigger mechanism, so that re-planning is too frequent or response is delayed, and the technical contradiction of computing resource consumption and motion continuity reduction is further aggravated. According to the invention, by constructing a closed-loop contro