CN-121979260-A - Unmanned aerial vehicle automatic analysis obstacle avoidance method

CN121979260ACN 121979260 ACN121979260 ACN 121979260ACN-121979260-A

Abstract

The invention discloses an automatic analysis obstacle avoidance method of an unmanned aerial vehicle, in particular relates to the technical field of autonomous flight control of the unmanned aerial vehicle, and aims to solve the technical problem that an obstacle avoidance safety boundary is unreliable due to scale drift of a visual map when the conventional unmanned aerial vehicle autonomously flies in long-term dependence on the historical map; the method is realized by judging task dependency, aligning histories with a real-time map and analyzing a reprojection error structure to quantify scale differences, further calculating mapping deviation of historic task tracks, comprehensively evaluating the influence degree of the current obstacle avoidance planning safety boundary, and finally adaptively carrying out space scale compensation on an environment map or a planning path according to the influence degree, thereby improving the obstacle avoidance decision reliability and safety of the unmanned aerial vehicle in long-term autonomous operation.

Inventors

XU ZELONG
ZHU JUNQUAN

Assignees

苏州慧行视界科技有限公司
广东环境保护工程职业学院

Dates

Publication Date: 20260505
Application Date: 20260128

Claims (10)

1. An unmanned aerial vehicle automatic analysis obstacle avoidance method is characterized by comprising the following steps: s1, an unmanned aerial vehicle constructs a real-time environment map through an airborne vision sensor; s2, judging whether the current flight mission is a long-term autonomous flight mission depending on a pre-stored historical environment map; S3, when the judgment is yes, aligning the real-time environment map with the historical environment map, and obtaining scale difference parameters by analyzing structural features of a re-projection error set generated by the same group of three-dimensional landmark points based on the historical pose of the historical environment map and the real-time pose of the real-time environment map respectively; S4, calculating behavior track deviation generated after mapping the historical typical task track recorded in the historical environment map to the real-time environment map based on the scale difference parameter; s5, evaluating the influence degree on the current obstacle avoidance path planning safety boundary by combining the scale difference parameter and the behavior track deviation; And S6, performing space scale compensation on the real-time environment map or the obstacle avoidance path generated based on the real-time environment map according to the influence degree.
2. The unmanned aerial vehicle automatic analysis obstacle avoidance method of claim 1, wherein the unmanned aerial vehicle constructs a real-time environment map through an on-board visual sensor, comprising: collecting a continuous image frame sequence of the environment through an airborne vision sensor; extracting and tracking visual feature points from the continuous image frame sequence to estimate pose change information of the unmanned aerial vehicle; And generating a three-dimensional point cloud representation of the real-time environment map through fusion processing based on the pose change information and the three-dimensional coordinates of the visual feature points.
3. The unmanned aerial vehicle automatic analysis obstacle avoidance method of claim 1, wherein determining whether the current flight mission is a long-term autonomous flight mission that relies on a pre-stored historical environmental map comprises: acquiring a task planning instruction of a current flight task, and analyzing indication information about whether a historical environment map is multiplexed in the task planning instruction; When the indication information requires multiplexing, further determining whether the current flight task is a long-term autonomous flight task depending on a pre-stored history environment map according to the flight path repetition execution period or the total duration of the task defined in the task planning instruction.
4. The unmanned aerial vehicle automatic analysis obstacle avoidance method of claim 1, wherein when the determination is yes, aligning the real-time environment map with the historical environment map, and obtaining the scale difference parameter by analyzing structural features of a re-projection error set generated by the same group of three-dimensional landmark points respectively based on the historical pose of the historical environment map and the real-time pose of the real-time environment map, comprises: Performing feature matching on the real-time environment map and the historical environment map, and calculating a space transformation relationship between the real-time environment map and the historical environment map to finish map alignment; Selecting a group of three-dimensional landmark points which are stably observed in the historical environment map and the real-time environment map as the same group of three-dimensional landmark points; based on the space transformation relation and the same group of three-dimensional landmark points, respectively calculating the reprojection errors of the three-dimensional landmark points under the historical pose and the real-time pose to form a reprojection error set; Extracting statistical distribution characteristics of the re-projection error set as structural characteristics; And calculating according to the structural characteristics to obtain the scale difference parameters.
5. The unmanned aerial vehicle automatic analysis obstacle avoidance method according to claim 4, wherein the step of extracting the statistical distribution characteristics of the re-projection error set as the structural characteristics comprises the steps of calculating a covariance matrix of the re-projection error set, carrying out eigenvalue decomposition on the covariance matrix to obtain eigenvalues and eigenvectors, and taking the distribution and ratio relation of main eigenvalues as the structural characteristics.
6. The unmanned aerial vehicle automatic analysis obstacle avoidance method of claim 1, wherein calculating a behavior track deviation generated after mapping a historical typical task track recorded in a historical environment map to a real-time environment map based on the scale difference parameter comprises: reading a path point sequence of a historical typical task track from a historical environment map; performing spatial scale transformation on the path point sequence by using the scale difference parameters to generate a mapping track mapped to a real-time environment map coordinate system; comparing the positions of the mapping track and the historical typical task track on the corresponding path points, and calculating the position offset; and based on the position offset of all the corresponding path points, calculating to obtain the behavior track deviation.
7. The method for automatically analyzing obstacle avoidance of the unmanned aerial vehicle according to claim 6, wherein reading the historical typical task trajectory from the historical environment map comprises selecting a complete trajectory of a primary task with highest obstacle avoidance success rate and higher smoothness of the flight trajectory than a preset threshold value from a plurality of historical flight tasks recorded by the historical environment map as the historical typical task trajectory.
8. The unmanned aerial vehicle automatic analysis obstacle avoidance method of claim 1, wherein the evaluating the degree of impact on the current obstacle avoidance path planning safety boundary by combining the scale difference parameter and the behavior track deviation comprises: comparing the scale difference parameter with a preset scale difference threshold value, and determining a quantized value of geometric safety boundary shrinkage or expansion caused by scale distortion; meanwhile, comparing the behavior track deviation with a preset track tolerance threshold value, and determining a path deviation risk level of the task execution layer; And integrating the variation of the geometric safety boundary and the path deviation risk level, and judging the influence degree on the current obstacle avoidance path planning safety boundary based on a predefined risk mapping rule.
9. The unmanned aerial vehicle automatic analysis obstacle avoidance method of claim 1, wherein performing spatial scale compensation on the real-time environment map or the obstacle avoidance path generated based thereon according to the degree of influence comprises: Determining a corresponding spatial scale compensation factor according to the influence degree; judging whether the influence degree exceeds a preset degree threshold, if not, selecting coordinate points of a obstacle avoidance path generated based on a real-time environment map to be currently executed, and performing path scale correction; if the three-dimensional point cloud coordinates are exceeded, the spatial scale compensation factors are applied to all three-dimensional point cloud coordinates of the real-time environment map, and (5) completing map reconstruction.
10. The unmanned aerial vehicle automatic analysis obstacle avoidance method according to claim 9, wherein determining the corresponding spatial scale compensation factor according to the influence degree comprises establishing a lookup table or a preset functional relation model taking a scale difference parameter and a behavior track deviation as input and taking the compensation factor as output, substituting the currently obtained scale difference parameter and the behavior track deviation into the lookup table or calculating to obtain the spatial scale compensation factor.

Description

Unmanned aerial vehicle automatic analysis obstacle avoidance method Technical Field The invention relates to the technical field of unmanned aerial vehicle autonomous flight control, in particular to an unmanned aerial vehicle automatic analysis obstacle avoidance method. Background In the field of unmanned aerial vehicle autonomous flight control, in order to realize automatic obstacle avoidance under a complex environment, the unmanned aerial vehicle can sense the environment in real time through an onboard vision sensor, and a three-dimensional map of the surrounding environment is constructed based on vision synchronous positioning and mapping technology, so that the unmanned aerial vehicle can perform autonomous positioning and navigation in an unknown or known environment without depending on external positioning signals. For long-term autonomous flight applications requiring repeated execution of tasks such as inspection and monitoring, the unmanned aerial vehicle relies on an environment map established by the preface flight as priori knowledge of the subsequent flight, and accordingly, path planning and obstacle avoidance decision are important ways for improving the working efficiency and autonomy. However, in a long-term and large-range autonomous flight task, the problem of inherent scale uncertainty and drift exists in a visually constructed environment map, so that global scale references of the environment map may be inconsistent in different flight tasks or different stages of the same task, when an unmanned aerial vehicle performs obstacle avoidance path planning according to a map with such scale distortion, the planned theoretical safety boundary and the safety boundary in a real physical environment generate unpredictable deviation, so that the long-term reliability and safety of obstacle avoidance decision are directly damaged, and the unmanned aerial vehicle faces potential collision risks when repeatedly autonomous operation is performed by using a historical map, so that the application of the unmanned aerial vehicle in a long-term autonomous operation scene requiring high reliability is limited. Disclosure of Invention In order to overcome the above-mentioned drawbacks of the prior art, the present invention provides an automatic analysis obstacle avoidance method for an unmanned aerial vehicle to solve the problems set forth in the above-mentioned background art. In order to achieve the above purpose, the present invention provides the following technical solutions: An unmanned aerial vehicle automatic analysis obstacle avoidance method comprises the following steps: s1, an unmanned aerial vehicle constructs a real-time environment map through an airborne vision sensor; s2, judging whether the current flight mission is a long-term autonomous flight mission depending on a pre-stored historical environment map; S3, when the judgment is yes, aligning the real-time environment map with the historical environment map, and obtaining scale difference parameters by analyzing structural features of a re-projection error set generated by the same group of three-dimensional landmark points based on the historical pose of the historical environment map and the real-time pose of the real-time environment map respectively; S4, calculating behavior track deviation generated after mapping the historical typical task track recorded in the historical environment map to the real-time environment map based on the scale difference parameter; s5, evaluating the influence degree on the current obstacle avoidance path planning safety boundary by combining the scale difference parameter and the behavior track deviation; And S6, performing space scale compensation on the real-time environment map or the obstacle avoidance path generated based on the real-time environment map according to the influence degree. Further, the unmanned aerial vehicle constructs a real-time environment map through an onboard vision sensor, comprising: collecting a continuous image frame sequence of the environment through an airborne vision sensor; extracting and tracking visual feature points from the continuous image frame sequence to estimate pose change information of the unmanned aerial vehicle; And generating a three-dimensional point cloud representation of the real-time environment map through fusion processing based on the pose change information and the three-dimensional coordinates of the visual feature points. Further, determining whether the current flight mission is a long-term autonomous flight mission that depends on a pre-stored historical environment map includes: acquiring a task planning instruction of a current flight task, and analyzing indication information about whether a historical environment map is multiplexed in the task planning instruction; When the indication information requires multiplexing, further determining whether the current flight task is a long-term autonomous flight task depending on a