CN-121806780-B - Method for realizing coordination scheduling of multiple robots in vehicle detection based on large model

CN121806780BCN 121806780 BCN121806780 BCN 121806780BCN-121806780-B

Abstract

The invention relates to the technical field of robot data monitoring, in particular to a method for realizing coordination scheduling of multiple robots for vehicle detection based on a large model, which comprises the steps of obtaining running time window and running speed data and background traffic log of a robot to be detected running to a current road section; the method comprises the steps of constructing a traffic speed grid graph of a current road section according to the distribution condition of the current road section and a running time window by combining a background traffic log, simulating running time distribution according to the running speed data and the traffic speed grid graph to obtain a motion deduction deviation degree, obtaining an identity verification result according to the angle deviation between the final position and the initial position of the robot to be detected in the current road section and the shooting position of monitoring equipment respectively and combining the motion deduction deviation degree, and carrying out coordinated scheduling on the robot to be detected. The invention provides a key identity reference for the coordinated scheduling of multiple robots, and effectively solves the problems of path conflict, task gear interruption and the like caused by identity loss.

Inventors

LI GUOHUANG
ZENG PENG
LIN SHUIXIANG
LI RENJIE
LIN CHENYI

Assignees

罗普特科技集团股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260309

Claims (9)

1. The method for realizing the coordination scheduling of the multiple robots for vehicle detection based on the large model is characterized by comprising the following steps: acquiring a running time window and running speed data of a robot to be detected running to a current road section, and forming a background traffic log by vehicle running records of the current road section in the running time window; According to the distribution condition of a current road section and the distribution condition of a running time window of a robot to be detected, combining with the space-time characteristics of a vehicle running record in a background traffic log, constructing a traffic speed grid graph of the current road section, wherein the traffic speed grid graph comprises the steps of uniformly dividing the total mileage of the current road section by taking the mileage of the current road section as a first dimension and the total time length of the running time window as a second dimension, and uniformly dividing the total time length of the running time window to construct a two-dimensional grid graph, wherein the running time window is a time period between the running-in time and the running-out time of the robot to be detected to the current road section; simulating the running time distribution of the robot to be detected on the current road section according to the running speed data of the robot to be detected and combining the running speed grid diagram to obtain the motion deduction deviation degree of the robot to be detected; And according to the angle deviation conditions between the final position and the initial position of the robot to be detected in the current road section and the shooting position of the monitoring equipment, combining the motion deduction deviation degree to obtain an identity verification result of the robot to be detected in the current road section, and carrying out coordinated scheduling on the robot to be detected.
2. The method for realizing coordinated scheduling of vehicle detection by multiple robots based on a large model according to claim 1, wherein the running speed data comprises the highest running speed, the highest acceleration and the highest deceleration of the robot to be detected; the vehicle travel record includes a travel time, a travel location, and a travel speed of each authenticated device.
3. The method for realizing the coordinated scheduling of the multiple robots for vehicle detection based on the large model according to claim 1, wherein the updating of the speed information of the unit grids in the two-dimensional grid map to obtain the traffic speed grid map of the current road section according to the running speed and the running time contained in the running record of the vehicle in the background traffic log and the speed limit information of the current road section specifically comprises the following steps: And taking the maximum running speed limited by the current road section as the speed information of the unit grids in the two-dimensional grid chart, determining the unit grid corresponding to the vehicle running record according to the running time and the mileage length of the running position of the vehicle running record in the current road section, and updating the speed information of the unit grid corresponding to the vehicle running record according to the minimum value between the running speed of the vehicle running record and the maximum running speed limited by the current road section to obtain the passing speed grid chart of the current road section.
4. The method for realizing coordination scheduling of multiple robots for vehicle detection based on a large model according to claim 3, wherein the method is characterized by simulating the running time distribution of the robots to be detected on the current road section according to the running speed data of the robots to be detected in combination with a traffic speed grid graph to obtain the motion deduction deviation degree of the robots to be detected, and specifically comprises the following steps: Initializing the iteration step number and the driving mileage of the robot to be detected in the current road section, and acquiring a preset time step and the simulation speed of the robot to be detected; updating the simulation speed of the robot to be detected according to the running speed data of the robot to be detected and the speed information of the unit grids in the current road section running speed grid chart at each preset time step, and calculating the running mileage of the robot to be detected at each preset time step; Iterating the running process of the robot to be detected on the current road section by using a preset time step, and stopping when the running distance is greater than or equal to the total mileage length of the current road section; And determining a gain coefficient based on the ratio of the unit grids with updated speed information in the passing speed grid diagram, and weighting the difference ratio between the simulated running time length of the robot to be detected and the time length of the running time window by using the gain coefficient to obtain the motion deduction deviation degree of the robot to be detected.
5. The method for realizing coordination scheduling of multiple robots for vehicle detection based on a large model according to claim 4, wherein updating the simulation speed of the robot to be detected according to the running speed data of the robot to be detected and the speed information of the unit grids in the current road section running speed grid chart, and calculating the running mileage of the robot to be detected at each preset time step specifically comprises the following steps: acquiring an initial speed of a robot to be detected when driving to a current road section; For any iteration process, calculating the theoretical driving distance of the robot to be detected under the preset time step based on the initial speed and the maximum acceleration of the robot to be detected; determining a unit grid range of a preset time step in a passing speed grid chart according to a theoretical running distance under the preset time step, and acquiring speed information in the unit grid range and a minimum value of the highest running speed of the robot to be detected as a simulation speed under the preset time step; Based on the initial speed and the simulation speed of the robot to be detected, calculating the simulation acceleration of the robot to be detected from the initial speed to the simulation speed, and based on the initial speed and the simulation speed of the robot to be detected, calculating the driving mileage of the robot to be detected under a preset time step and the updated simulation speed.
6. The method for realizing coordination scheduling of multiple robots for vehicle detection based on a large model according to claim 1, wherein the obtaining the identity verification result of the robot to be detected on the current road section by combining the motion deduction deviation degree according to the angle deviation condition between the final position and the initial position of the robot to be detected on the current road section and the shooting position of the monitoring device, specifically comprises: Obtaining the visual weight of the robot to be detected on the current road section according to the deviation condition between the driving direction of the robot to be detected on the ending position of the current road section and the driving direction of the initial position of the robot to be detected; acquiring cosine similarity between an image feature vector of an image shot by the robot to be detected at the ending position of the current road section and an image feature vector of an image shot by the robot to be detected at the initial position, and taking the cosine similarity as a visual similarity index of the robot to be detected at the current road section; The visual weight is used for fusing the visual similarity index of the robot to be detected and the motion deduction deviation degree to obtain the verification confidence coefficient of the robot to be detected on the current road section; And acquiring an identity verification result of the robot to be detected on the current road section according to the verification confidence coefficient of the robot to be detected on the current road section and a preset confidence threshold.
7. The method for realizing coordinated scheduling of multiple robots for vehicle detection based on a large model according to claim 6, wherein the obtaining the visual weight of the robot to be detected on the current road section according to the deviation between the respective corresponding observation angles of the driving direction of the robot to be detected on the ending position of the current road section and the driving direction of the initial position specifically comprises: based on the ending position and the initial position of the robot to be detected at the current road section, respectively determining a lane tangential vector of the robot to be detected at the ending position of the current road section and a lane tangential vector of the robot to be detected at the initial position of the current road section, and acquiring a preset optical axis orientation vector of the detection system; and calculating an included angle between each lane tangential vector and the optical axis orientation vector as an observation angle corresponding to each lane tangential vector, and taking the cosine value absolute value of the difference absolute value between the observation angles corresponding to the end position and the initial position of the robot to be detected in the current road section as the visual weight of the robot to be detected in the current road section.
8. The method for realizing the coordinated scheduling of the multiple robots for vehicle detection based on the large model according to claim 6, wherein the method for obtaining the verification confidence of the robot to be detected in the current road section by fusing the visual similarity index and the motion deduction deviation degree of the robot to be detected by using the visual weight specifically comprises the following steps: Carrying out weighted summation on the negative correlation coefficient of the motion deduction deviation degree of the robot to be detected and the visual similarity index to obtain the verification confidence coefficient of the robot to be detected on the current road section; The weight corresponding to the negative correlation coefficient of the motion deduction deviation degree is the negative correlation coefficient of the visual weight, and the weight corresponding to the visual similarity index is the visual weight.
9. The method for realizing coordination scheduling of multiple robots for vehicle detection based on a large model according to claim 6, wherein the step of obtaining the identity verification result of the robot to be detected on the current road section according to the verification confidence of the robot to be detected on the current road section and a preset confidence threshold value specifically comprises the following steps: When the verification confidence coefficient of the robot to be detected in the current road section is greater than or equal to a preset confidence threshold value, the robot to be detected passes identity verification; When the verification confidence coefficient of the robot to be detected on the current road section is smaller than a preset confidence threshold value, the robot to be detected does not pass the identity verification.

Description

Method for realizing coordination scheduling of multiple robots in vehicle detection based on large model Technical Field The invention relates to the technical field of robot data monitoring, in particular to a method for realizing coordination scheduling of multiple robots for vehicle detection based on a large model. Background In scenes such as closed commodity circulation garden, automatic harbour, unmanned collection card's robot's cooperation dispatch, the core premise is that the continuous accurate discernment of realization vehicle identity. The key nodes in the operation area are deployed with road side sensing units (such as cameras and millimeter wave radars), but are affected by wide field range and more shielding objects, the vision fields of the sensing units cannot completely cover the whole areas, and uncontrolled blind areas are formed between the observation points. After the vehicle passes through the blind area, when entering the next observation point from different angles, the visual appearance characteristics can be obviously drifted due to the severe transformation (such as front view to side view) of the shooting visual angle, and the problem directly causes the failure of the traditional visual re-recognition algorithm. On the basis, uncertain factors such as networking-free social vehicles, temporary obstacles and the like exist in the blind area, the road condition height is time-varying, the actual passing duration of the vehicles is in nonlinear fluctuation, and the traditional prediction method based on static distance or average speed cannot distinguish whether the fluctuation is caused by road condition congestion or vehicle self-performance limitation. The visual recognition failure superposition movement duration prediction misalignment can cause that a system cannot judge whether a cross-vision front and back observation record belongs to the same vehicle or not, further the vehicle identity continuity is lost, the scheduling context such as task priority, road right lock and the like is caused to be disjointed, and finally, the scheduling confusion such as robot path conflict, task repeated allocation and the like is caused, so that the operation efficiency and stability are seriously influenced. Disclosure of Invention In order to solve the technical problem that the accuracy of the identity verification result of the cross-vision robot vehicle by the existing method is poor, the invention aims to provide a method for realizing the coordination scheduling of multiple robots for vehicle detection based on a large model, and the adopted technical scheme is as follows: acquiring a running time window and running speed data of a robot to be detected running to a current road section, and forming a background traffic log by vehicle running records of the current road section in the running time window; According to the distribution condition of the current road section and the distribution condition of the running time window of the robot to be detected, combining the space-time characteristics of the vehicle running record in the background traffic log to construct a traffic speed grid diagram of the current road section; simulating the running time distribution of the robot to be detected on the current road section according to the running speed data of the robot to be detected and combining the running speed grid diagram to obtain the motion deduction deviation degree of the robot to be detected; And according to the angle deviation conditions between the final position and the initial position of the robot to be detected in the current road section and the shooting position of the monitoring equipment, combining the motion deduction deviation degree to obtain an identity verification result of the robot to be detected in the current road section, and carrying out coordinated scheduling on the robot to be detected. Preferably, the running time window is a time period between the running-in time and the running-out time of the robot to be detected running to the current road section; The running speed data comprise the highest running speed, the highest acceleration and the highest deceleration of the robot to be detected; the vehicle travel record includes a travel time, a travel location, and a travel speed of each authenticated device. Preferably, the constructing a traffic speed grid chart of the current road section according to the distribution situation of the current road section and the distribution situation of the running time window of the robot to be detected in combination with the space-time characteristics of the vehicle running record in the background traffic log specifically includes: taking the mileage length of the current road section as a first dimension, taking the total time length of the running time window as a second dimension, uniformly dividing the total mileage length of the current road section, uniformly dividing the total time length of