CN-121982718-A - Intelligent logistics distribution method based on multi-mode logistics knowledge graph

CN121982718ACN 121982718 ACN121982718 ACN 121982718ACN-121982718-A

Abstract

The invention relates to the field of logistics distribution environment perception, in particular to an intelligent logistics distribution method based on a multi-mode logistics knowledge graph. The method comprises the steps of obtaining road images through a visual sensor, constructing a visual mutation observation range, selecting a visual characteristic map to extract corresponding visual characteristics in the visual mutation observation range, constructing a time domain change curve of the extracted visual characteristics, and dynamically capturing a sensitive time domain and a non-sensitive time domain in a driving environment according to the real-time change condition of the curve. And in the non-sensitive time domain, converting into a mode of determining whether a target for visual detection has a corresponding point cloud cluster at intervals. Therefore, the new energy logistics vehicle can strengthen the perceived reliability in a complex high-risk period, and the perceived reliability of the system, the overall energy efficiency and the resource utilization rate are improved.

Inventors

DONG YINDI
PENG WEI
TANG YU
JIANG LIHUA
DENG CHANGCHUN

Assignees

重庆城市管理职业学院

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. An intelligent logistics distribution method based on a multi-mode logistics knowledge graph is characterized by comprising the steps of, Acquiring image data of visual sensors distributed by the logistics distribution unit, and marking boundary features according to the image data; constructing a visual mutation observation range according to the boundary features, and selecting a visual feature map to extract corresponding visual features in the visual mutation observation range, wherein the visual feature map comprises route types of a plurality of route segments and visual features associated with the route types; constructing a visual characteristic time domain change curve according to visual characteristics extracted from continuous image data, and capturing a sensitive time domain according to the change condition of the visual characteristic time domain change curve; radar data acquired by a laser radar distributed by a logistics distribution unit is acquired, and the radar data is analyzed according to a sensitive time domain and a non-sensitive time domain to synchronously analyze a judging result aiming at image data, including, Analyzing radar data in real time, acquiring geometric features of the laser radar, constructing a geometric feature time domain change curve, and comparing the change trend of the geometric feature time domain change curve with the change trend of the geometric feature time domain change curve to judge whether the image data is abnormal or not; analyzing the radar data at intervals, and verifying whether the target features in the image data have corresponding point cloud clusters or not so as to judge whether the image data are abnormal or not.
2. The intelligent logistics distribution method based on the multi-modal logistics knowledge graph as set forth in claim 1, wherein the process of labeling boundary features based on the image data comprises, And identifying the boundary of the passable road as a boundary characteristic, and marking.
3. The intelligent logistics distribution method based on multi-modal logistics knowledge graph as set forth in claim 1, wherein the process of constructing the visual mutation observation range according to the boundary features includes, A strip-shaped area with preset width is determined along the extending direction of the passable road at the inner side of the boundary feature of the passable road; the banded region is determined as a visual range of review of the variability.
4. The intelligent logistics distribution method based on the multi-modal logistics knowledge graph as claimed in claim 1, wherein the visual characteristic graph comprises route types of a plurality of route segments, wherein the route types are divided into high dynamic road segments and low dynamic road segments based on real-time traffic flow, If the real-time traffic flow is greater than or equal to a preset flow threshold, dividing into a high-dynamic road section; If the real-time traffic flow is smaller than the preset flow threshold, the real-time traffic flow is divided into low-dynamic road sections.
5. The method for intelligent logistics distribution based on multi-modal logistics knowledge map as set forth in claim 1, wherein the visual characteristics associated with the route types include, If the road sections are divided into high dynamic road sections, the associated visual characteristics are the pixel duty ratio of the non-road-surface chromaticity range in the visual variation observation range; if the road segments are divided into low dynamic road segments, the associated visual characteristic is the average chromaticity in the visual mutation observation range.
6. The intelligent logistics distribution method based on the multi-modal logistics knowledge graph as set forth in claim 5, wherein the capturing the sensitive time domain according to the change condition of the visual characteristic time domain change curve comprises, If the pixel duty ratio of the non-road chromaticity range is greater than or equal to a preset pixel duty ratio threshold value or the average chromaticity is greater than or equal to a preset chromaticity threshold value, bidirectionally extending the corresponding moment for a preset time length, and capturing the extended time domain segment as a sensitive time domain; The time period in which the visual characteristic time domain change curve corresponds to the time domain that is not captured as the sensitive time domain is determined as the non-sensitive time domain.
7. The intelligent logistics distribution method based on multi-modal logistics knowledge graph as set forth in claim 1, wherein the synchronous analysis of the determination result for the image data comprises, If the capture is a sensitive time domain, analyzing the radar data in real time to obtain the geometric features of the laser radar, constructing a geometric feature time domain change curve, and comparing the change trend of the geometric feature time domain change curve with the change trend of the geometric feature time domain change curve to judge whether the image data is abnormal or not; If the time domain is determined to be a non-sensitive time domain, analyzing the radar data at intervals, and verifying whether the target features in the image data have corresponding point cloud clusters or not so as to judge whether the image data are abnormal or not.
8. The method for intelligent logistics distribution based on multi-modal logistics knowledge graph as set forth in claim 7, wherein the process of comparing the trend of the visual characteristic time domain change curve with the trend of the geometric characteristic time domain change curve includes, Determining the volume change rate of the point cloud cluster in the visual mutation observation range, and constructing a geometrical characteristic time domain change curve based on the volume change rate; and aligning the visual characteristic time domain change curve and the geometric characteristic time domain change curve in time domain, and comparing the change trend and the change slope in each time domain segment.
9. The intelligent logistics distribution method based on multi-modal logistics knowledge graph as set forth in claim 8, wherein the comparing the trend of the visual characteristic time domain change curve with the geometric characteristic time domain change curve to determine whether the image data is abnormal comprises, If the change trends are inconsistent and the difference ratio of the change slopes is smaller than a preset difference ratio threshold, the image data is judged to be abnormal.
10. The intelligent logistics distribution method based on multi-modal logistics knowledge map as set forth in claim 7, wherein verifying whether the target feature in the image data has a corresponding point cloud cluster to determine whether the image data is abnormal comprises, If the target feature in the image data does not have the corresponding point cloud cluster, judging that the image data is abnormal.

Description

Intelligent logistics distribution method based on multi-mode logistics knowledge graph Technical Field The invention relates to the field of logistics distribution environment perception, in particular to an intelligent logistics distribution method based on a multi-mode logistics knowledge graph. Background Along with the deep fusion of the Internet of things, big data and artificial intelligence technology, intelligent logistics has become a core driving force for improving the efficiency of a supply chain and reducing the operation cost. Under the current background of 'double carbon' target and energy structure transformation, electric delivery vehicles represented by new energy logistics vehicles are being popularized rapidly, and the endurance, energy efficiency optimization and operation reliability of the electric delivery vehicles provide higher requirements on the instantaneity and energy efficiency of a sensing system. However, complex and changeable urban distribution environments, such as abrupt traffic flow, weather influence, mixed road right scenes and the like, bring serious challenges to the environment sensing method based on a single mode, namely, continuous high-load full-time domain high-precision sensing can improve safety, but the electricity consumption and calculation burden of new energy vehicles can be obviously increased, the endurance mileage and the system economy are influenced, and the sensing reliability is difficult to ensure under the sudden risk scene by a simple timing or low-power consumption sensing strategy. Therefore, how to realize the dynamic optimization of perceived resources and energy consumption while ensuring the perceived accuracy of environment becomes a key problem to be solved in the large-scale and intelligent development of new energy logistics motorcades. CN119225366a discloses a method and a device for detecting unmanned vehicle obstacle in open-air mining area. The method comprises the steps of receiving pose data detected by inertial navigation systems arranged on four positions of an unmanned vehicle in an open pit mining area, wherein the four positions correspond to four wheels, at least one inertial navigation system is arranged on any position, determining that at least one of the wheels corresponding to any two positions rolls a protrusion or a pit under the condition that the deviation of the pose data detected on any two positions is larger than a preset deviation, and determining a height abnormal wheel according to the height data of the wheels, wherein the deviation of the pose data comprises position deviation and angle deviation, the height data is obtained from the pose data, the height abnormal wheel is a wheel with abnormal height data, and determining the position information and the size information of the protrusion or the pit based on the pose data of the height abnormal wheel. The scheme solves the problem that the perception capability of the obstacle detection method of the unmanned vehicles in the open-air mining area in the prior art is limited. There are problems in the prior art that, In the autonomous distribution process of the unmanned vehicle, the visual perception is easy to be interfered by the environment to generate false alarms, and the system is usually required to continuously perform high-load fusion verification, so that not only can the erroneous judgment of obstacles be possibly caused, but also huge redundant calculation can be generated in the long-period safe running, thereby reducing the perceived reliability of the system, the overall energy efficiency and the resource utilization rate. Disclosure of Invention Therefore, the invention provides an intelligent logistics distribution method based on a multi-mode logistics knowledge graph, which is used for overcoming the problems that in the prior art, false alarms are generated due to the fact that visual perception is easy to be interfered by environment, and a system is usually required to continuously perform high-load fusion verification, so that not only is the erroneous judgment of obstacles possibly caused, but also huge redundant calculation is possibly generated in long-term safe driving, and therefore the perceived reliability of the system and the overall energy efficiency and resource utilization rate are reduced. In order to achieve the above object, the present invention provides an intelligent logistics distribution method based on a multi-modal logistics knowledge graph, which comprises, Step S1, obtaining image data of visual sensors distributed by a logistics distribution unit, and marking boundary features according to the image data; step S2, constructing a visual mutation observation range according to the boundary features, and selecting a visual feature map to extract corresponding visual features in the visual mutation observation range, wherein the visual feature map comprises route types of a plurality of route segments and vis