Operation control method and system of unmanned transport truck
By acquiring cargo constraints, multi-dimensional feature acquisition and real-time monitoring, combined with predetermined paths and operating modes, the problem of low operation control accuracy and stability of unmanned trucks is solved, and high-precision, efficient and safe operation control is achieved.
Patent Information
- Application Number
- CN202311314070.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-11
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2043-10-11
AI Technical Summary
The existing unmanned truck operation control accuracy and stability are low, mainly due to the inaccurate position and motion state monitoring, and the single trajectory planning and control strategy.
By acquiring the cargo constraints of unmanned trucks in the operation area, multi-dimensional feature acquisition is carried out to draw a grid map, analyze the predetermined target path, and perform high-precision control in combination with real-time operation mode, using position sensors and CCD image sensors for real-time use in real-time Monitoring and closed-loop control ensure efficient and safe transportation.
It improves the control accuracy and stability of unmanned trucks, achieves high-precision, efficient and safe operation control, and enhances the real-time monitoring and adaptability of trucks.
Smart Images

Figure CN117170386B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent control, and in particular to an operation control method and system for an unmanned truck. Background Art
[0002] With the development of artificial intelligence, machine vision and mobile Internet technology, unmanned trucks have been widely used in various industries. However, the current monitoring of the position and motion status of unmanned trucks is not accurate enough, resulting in low operation control accuracy; the implementation of simple trajectory planning and a single control strategy for trucks leads to low operation stability. Summary of the invention
[0003] The present application provides an operation control method and system for an unmanned truck, aiming to solve the technical problem of low operation control accuracy and stability of unmanned trucks in the prior art.
[0004] In view of the above problems, the present application provides an operation control method and system for an unmanned truck.
[0005] The first aspect disclosed in the present application provides an operation control method for an unmanned truck, the method comprising: obtaining target freight constraints of a target unmanned truck in a target operation area, wherein the target freight constraints include a freight starting point and a freight end point; performing multi-dimensional feature acquisition on the target operation area to obtain target area information, and drawing a target grid map based on the target area information; analyzing and obtaining a predetermined target path based on the freight starting point, the freight end point and the target grid map, wherein the predetermined target path refers to a plurality of location points with turning angle identifiers; obtaining a real-time operation mode of the target unmanned truck; and in the real-time operation mode, performing operation control on the target unmanned truck in combination with the predetermined target path.
[0006] Another aspect disclosed in the present application provides an operation control system for an unmanned truck, the system comprising: a target freight constraint module, used to obtain target freight constraints of a target unmanned truck in a target operation area, wherein the target freight constraints include a freight starting point and a freight end point; an area feature acquisition module, used to perform multi-dimensional feature acquisition on a target operation area to obtain target area information, and draw a target grid map based on the target area information; a predetermined path analysis module, used to analyze and obtain a predetermined target path based on the freight starting point, the freight end point and the target grid map, wherein the predetermined target path refers to a plurality of location points with steering angle identifiers; an operation mode acquisition module, used to obtain a real-time operation mode of a target unmanned truck; and a target operation control module, used to perform operation control of the target unmanned truck in combination with the predetermined target path under the real-time operation mode.
[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:
[0008] The method adopts the method of obtaining the cargo transportation constraints of the unmanned truck in the target operation area, including the starting point and end point of the cargo transportation; obtaining the target area information by multi-dimensional feature collection of the target operation area, and drawing the target grid map based on the target area information; then obtaining the predetermined target path according to the cargo transportation starting point, end point and target grid map analysis; and obtaining the real-time operation mode of the unmanned truck at the same time; and under the real-time operation mode, controlling the operation of the unmanned truck in combination with the predetermined target path, solves the technical problem of low operation control accuracy and stability of the unmanned truck in the prior art, and achieves the technical effect of improving the control accuracy and stability of the unmanned truck.
[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] Figure 1 A possible flow chart of an operation control method of an unmanned truck is provided for an embodiment of the present application;
[0011] Figure 2 A flow chart of a communication connection of a job control system is provided for an embodiment of the present application;
[0012] Figure 3 A possible structural schematic diagram of an operation control system of an unmanned truck is provided for an embodiment of the present application.
[0013] Explanation of the reference numerals: target freight constraint module 11, regional feature acquisition module 12, predetermined path analysis module 13, operation mode acquisition module 14, target operation control module 15. DETAILED DESCRIPTION
[0014] The overall idea of the technical solution provided by this application is as follows:
[0015] The embodiment of the present application provides an operation control method and system for an unmanned truck, which adopts multimodal perception and adaptive control strategies to achieve high-precision, high-efficiency and highly adaptable operation control. First, the target area information is obtained by stereoscopically collecting the target operation area, and the target grid map is drawn based on the target area information. The predetermined target path is obtained in combination with the unmanned truck cargo constraint analysis to provide a basis for subsequent operation control; at the same time, the real-time operation mode of the unmanned truck is obtained, and an appropriate operation control strategy is selected; during the operation control process, the real-time position information of the unmanned truck is obtained through monitoring by the position sensor, the target CCD image sensor and the intelligent recognition model, and compared with the predetermined target path to obtain the real-time steering deviation angle, so as to achieve high-precision position closed-loop control. In addition, the real-time quality of the unmanned truck is monitored by the quality sensor installed on the unmanned truck, and the integrity of the transported goods is detected.
[0016] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.
[0017] Embodiment 1
[0018] like Figure 1 As shown, an embodiment of the present application provides an operation control method for an unmanned truck. The operation control method is applied to an operation control system, and the operation control system is communicatively connected with a target unmanned truck.
[0019] Specifically, an embodiment of the present application provides an operation control method for an unmanned truck. The method is applied to an operation control system, and the operation control system establishes a communication connection with the unmanned truck.
[0020] The operation control system includes a control device for generating operation control instructions and a communication device for communicating with the unmanned truck. The control device generates appropriate operation control instructions based on the environmental information, regional information and the operation information of the unmanned truck itself acquired by various sensors, and sends them to the unmanned truck through the communication device, so that it can automatically and stably operate and transport.
[0021] The communication connection between the operation control system and the unmanned truck can be established based on wired communication or wireless communication. Among them, wired communication methods include RS-232, RS-485 and Control Area Network, and wireless communication methods include WiFi, Bluetooth, ZigBee and mobile communication. Through the communication connection, the operation control system can monitor the status of the unmanned truck in real time and send control instructions to the unmanned truck when necessary, so as to realize real-time monitoring and precise control of the unmanned truck, so that the unmanned truck can operate and transport automatically, efficiently and safely.
[0022] Job control methods include:
[0023] Step S1000: Obtaining target freight constraints of the target unmanned truck in a target operation area, wherein the target freight constraints include a freight start point and a freight end point;
[0024] Specifically, the starting point and end point of the target unmanned truck in the target operation area are obtained based on the pre-planned transportation task data or the detection results of the environmental monitoring module. Among them, the pre-planned transportation task data refers to the detailed information of various transportation tasks that the unmanned truck needs to perform, including the data of the starting point and the end point, which are pre-stored in the operation control system. After the unmanned truck receives the instruction to perform a certain task, it directly extracts the corresponding starting point and end point information from the stored data; the detection of the environmental monitoring module is to detect the environment where the drone is located through various environmental monitoring devices deployed by the operation control system, such as cameras, laser radars, and RFID readers and writers. When the monitoring equipment detects a target that meets the starting point or end point conditions, the data it reads is used as the information of the transportation constraint. For example, the camera identifies the area that meets the loading conditions, and the RFID reader and writer reads the specified RFID tag, etc.
[0025] By obtaining the target cargo transportation constraints of the target unmanned truck in the target operating area, the necessary reference information is provided for subsequent path planning and motion control, enabling the unmanned truck to operate transportation accurately and efficiently.
[0026] Step S2000: performing multi-dimensional feature acquisition on the target operation area to obtain target area information, and drawing a target grid map based on the target area information;
[0027] Specifically, the target operation area refers to the work area where the unmanned truck needs to operate and transport. Environmental monitoring equipment is deployed in this area to monitor the operating status and environmental conditions of the unmanned truck. Multi-dimensional feature acquisition includes obtaining spatial geometric information, obstacle distribution information, surface material information, etc. of the target operation area.
[0028] First, a variety of sensors, such as cameras, laser radars, and RFID readers, are arranged in the target operation area to obtain multi-dimensional information in the area, such as spatial geometry, obstacle distribution, and surface friction. Secondly, a SLAM system is formed using visual sensors and inertial measurement units. When the SLAM system moves in the target operation area, it synchronously collects image data and its own motion data, and reconstructs a three-dimensional map of the area in real time through the SLAM algorithm. Then, based on the three-dimensional digital map, a three-dimensional grid division algorithm is used to divide the continuous three-dimensional space into multiple small areas, each area as a grid unit, and each grid unit is given the information represented by the space, such as obstacle level, surface friction coefficient, etc. These grid information ultimately constitute the target grid map. By obtaining detailed environmental information of the target operation area and drawing the target grid map, the foundation for the automatic operation control of unmanned trucks is laid.
[0029] Step S3000: Analyze and obtain a predetermined target path according to the cargo transportation starting point, the cargo transportation destination and the target grid map, wherein the predetermined target path refers to a plurality of position points with turning angle identifiers;
[0030] Specifically, the operation control system uses the path planning algorithm to analyze and calculate the predetermined target path of the unmanned truck based on the acquired cargo starting and ending information and the target grid map. The predetermined target path consists of multiple location points, each of which identifies the angle at which the unmanned truck needs to turn. These paths and steering information can accurately guide the unmanned truck from the starting point to the end point.
[0031] First, the operation control system extracts the coordinates of the unmanned truck's starting and ending points and the corresponding grid numbers from the stored data or the detection results of the environmental monitoring module; at the same time, the stored target grid map is read to parse the information of each grid unit, such as obstacle level, surface friction coefficient, etc., for judging the feasibility of the grid in the path planning calculation; then, the A* algorithm is used as the path planning algorithm, and the search space range of the path planning is determined according to the starting and ending grids, and only the grid network between the starting and ending points is considered; then, in the search space, the estimated moving cost from each grid to the ending grid is calculated and recorded in the open list in the search space, and the estimated cost of the starting grid is set to 0; then, the grid with the smallest estimated cost in the open list is selected as the extended grid, and is deleted from the open list, and the path from the extended grid to the starting grid is recorded; then, the adjacent grids of the extended grid are extended, and the estimated cost from the adjacent grids to the ending point is calculated and recorded in the open list, and the extended grid is repeatedly selected until the extended grid is the ending grid; finally, the recorded path is the best path from the starting point to the ending grid, which is the predetermined target path.
[0032] The predetermined target path of the unmanned truck is obtained through path planning algorithm analysis, providing the necessary route reference information for its subsequent automatic navigation and control, guiding the unmanned truck to complete the transportation task safely and efficiently.
[0033] Step S4000: obtaining the real-time operation mode of the target unmanned truck;
[0034] Specifically, the operation control system obtains the current operation mode of the unmanned truck, which mainly includes remote control driving mode, automatic driving mode and remote driving mode. The operation control system can actively select the appropriate operation mode according to the task requirements and operating environment. For example, the remote control mode is selected in areas with complex sections or many obstacles, and the automatic driving mode is selected in simple sections. By obtaining the current operation mode of the unmanned truck, it provides an important reference for the operation control system to generate corresponding control instructions and adjust the control strategy.
[0035] Among them, in the remote control driving mode, the operating status and driving path of the unmanned truck are controlled by a manual follow-up truck through a handle control device, and control instructions are sent to the unmanned truck to achieve control of its speed, direction, etc.; in the automatic driving mode, the unmanned truck automatically drives to the destination according to the pre-stored driving route and control instructions provided by the operation control system. In this mode, the unmanned truck has the ability of automatic driving and control, and can implement speed control, steering control and braking control according to the path information; in the remote driving mode, the operator sends control instructions to the unmanned truck through the remote control terminal to achieve remote control of its status and movement. In this mode, the unmanned truck and the remote control terminal receive and respond to control instructions in real time through a communication connection.
[0036] Step S5000: In the real-time operation mode, the target unmanned truck is controlled in operation in combination with the predetermined target path.
[0037] Specifically, the operation control system generates corresponding control instructions based on the real-time operation mode of the unmanned truck and the predetermined target path, and performs operation control on the unmanned truck. For example, in the manual control mode, the operation control system provides the predetermined target path information to the manual operator as a navigation reference, and the manually sent control instructions are directly transmitted to the unmanned truck to achieve manual driving control; in the automatic driving mode, the operation control system generates an automatic navigation path based on the predetermined target path, and generates control instructions, which are sent to the unmanned truck to achieve automatic driving control; in the remote control mode, the operation control system transmits the predetermined target path information to the remote control terminal for the operator's reference, and at the same time directly sends the control instructions generated manually at the remote terminal to the unmanned truck to achieve remote control.
[0038] By adopting corresponding control strategies according to different operating modes and implementing precise control in combination with the predetermined target path, unmanned trucks can complete navigation operations automatically or manually, achieve high-precision, efficient and safe operation of unmanned trucks, and provide basic guarantees for the completion of transportation tasks.
[0039] Further, such as Figure 2 As shown, the embodiment of the present application also includes:
[0040] Step S2100: acquiring basic target information of the target operation area, wherein the basic target information includes shape data and size data;
[0041] Step S2200: establishing an initial model of the target operation area according to the shape data and the size data;
[0042] Step S2300: acquiring target layout information of the target operation area, and rendering the target layout information to the initial model to obtain a target model;
[0043] Step S2400: performing grid division on the target model to obtain a grid division result, wherein the grid division result includes a plurality of grids;
[0044] Step S2500: performing judgment and analysis on each of the plurality of grids in turn, and obtaining the target grid map according to the analysis results.
[0045] Specifically, first, three-dimensional sensors such as cameras and lidar are deployed in the target operation area to collect three-dimensional spatial information of the area; the three-dimensional sensors scan the target operation area to obtain three-dimensional point cloud data in the area to describe the three-dimensional geometric boundary shape of the area; the operation control system receives the three-dimensional point cloud data and filters and aligns the data. The processed three-dimensional point cloud data expresses the basic geometric shape of the area and provides dimensional parameter information; the shape and size information is used as the basic information of the target operation area and stored in the operation control system to provide an initial data source for modeling processing.
[0046] Secondly, the operation control system uses 3D mesh technology to establish an initial 3D model of the target operation area with the basic information obtained as input. The initial 3D model expresses the basic geometric outline of the target operation area, but does not contain detailed internal environmental information, providing a reference for the fusion of environmental information and the establishment of the target model. Next, the environmental monitoring module collects detailed environmental information in the target operation area, such as internal road structure, obstacle distribution, and surface condition information. The 3D sensor scans the target operation area again to obtain the internal environmental 3D point cloud data, which integrates the 3D position information of various environmental elements. Subsequently, the operation control system registers and fuses the environmental 3D point cloud data with the initial 3D model, uses 3D reconstruction technology to render the environmental information into the model, and generates a target 3D model of the target operation area. The target 3D model contains the fine 3D geometric structure information and internal detailed environmental information of the target operation area, which is stored in the operation control system.
[0047] Then, the operation control system performs three-dimensional grid division on the stored target three-dimensional model, and uses a three-dimensional division algorithm to divide the continuous three-dimensional model space into multiple small units, each unit is called a grid, and determines the spatial range, volume size and three-dimensional coordinate information of each grid, and the division results are multiple grids and their related parameters. The grid division results are stored in the operation control system to provide data for subsequent grid judgment and analysis. Then, the operation control system extracts the obtained multiple grid information in turn, and judges and analyzes the environmental characteristics inside each grid. According to the preset judgment conditions, such as the number of obstacles, traffic capacity and surface conditions, it is judged whether each grid meets the requirements, and each grid is marked whether it can be driven. Subsequently, all grid judgment results are formed into a target grid map according to the identification results, providing the necessary path planning basis for the automatic operation control of the unmanned truck.
[0048] Furthermore, the embodiment of the present application also includes:
[0049] Step S2510: extracting any one grid from the plurality of grids and recording it as a target grid;
[0050] Step S2520: determining whether the target grid meets the predetermined requirements, if so, performing a first mark on the target grid, and if not, performing a second mark on the target grid;
[0051] Step S2530: construct the target grid map based on the first mark or the second mark.
[0052] Specifically, first, a grid is randomly selected from the multiple grids obtained and recorded as the target grid to be judged. Then, the environmental feature information inside the target grid is judged and analyzed to determine whether it meets the pre-set conditions, such as the number of obstacles, the surface friction coefficient, and the width of the passage. If the environmental characteristics in the target grid meet the requirements, it is judged as the first mark, and if it does not meet the requirements, it is judged as the second mark. Among them, the first mark indicates that the target grid is suitable for unmanned trucks to travel and operate, and the second mark indicates that it is not suitable for driving. Subsequently, according to the first mark or the second mark result of the target grid, the target grid is added to the target grid map, and the target grid map is finally composed of the judgment marks of all grids.
[0053] In the target grid map, the grid with the first mark is used as an optional grid for the path planning of the unmanned truck, forming its navigation network. The grid with the second mark needs to be avoided. The path generated by the operation control system based on the target grid map and the path planning algorithm will only pass through the grid with the first mark. By judging and analyzing the environmental characteristics of each grid, the first mark or the second mark is obtained, and finally the target grid map is constructed in the form of judgment marks, which expresses the feasibility information of the internal environment navigation of the target operation area and provides a reference for the path planning and control of the unmanned truck.
[0054] Furthermore, the embodiment of the present application also includes:
[0055] Step S3100: matching the first grid of the cargo transportation starting point and the second grid of the cargo transportation destination respectively;
[0056] Step S3200: reading a predetermined traversal scheme, and obtaining a first adjacent grid set of the first grid based on the predetermined traversal scheme, wherein the first adjacent grid set includes M adjacent grids, and M is an integer greater than or equal to 1;
[0057] Step S3300: extracting a first adjacent grid from the M adjacent grids, and calculating a first estimated cost from the first adjacent grid to the second grid;
[0058] Step S3400: selecting the first adjacent grid of the target with the first estimated cost being minimized as a constraint;
[0059] Step S3500: taking the first grid as a first position point, combining the first grid and the target first adjacent grid to obtain a first position steering angle;
[0060] Step S3600: Obtain the predetermined target path according to the first position point and its first position steering angle.
[0061] Specifically, first search for the grid corresponding to the starting point of the shipment in the target grid map, recorded as the first grid. Similarly, find the destination grid, recorded as the second grid. The obtained first grid and second grid are used as the starting and ending nodes of the A* path search. Various grid traversal schemes are pre-stored in the job control system, such as breadth-first, shortest path first, etc. Read the pre-traversal scheme as the strategy for this path search. According to the pre-traversal scheme and the position of the first grid, find all adjacent grids connected to the first grid. These adjacent grids constitute the first adjacent grid set, which contains M adjacent grids, and M is an integer greater than or equal to 1. The adjacent grid set provides optional directions and ranges for searching starting from the first grid.
[0062] Then, a neighboring grid is selected from the first neighboring grid set as the first neighboring grid, and the estimated cost d from the neighboring grid to the second grid (the end grid) is calculated. The estimated cost is half of the sum of the Euclidean distance d′ and the Manhattan distance d″, that is, As the evaluation criterion of A* search, it is used to select the search path. The smaller the cost, the better the path. Then, the estimated cost of each adjacent grid to the second grid in the first adjacent grid set is compared, and the adjacent grid with the smallest estimated cost is selected as the target first adjacent grid.
[0063] Subsequently, the first grid is set as the first position point of the path, and the first position steering angle is calculated according to the relative orientation relationship between the first grid and the target first adjacent grid. The steering angle indicates the angle at which the unmanned truck needs to turn from the first position point to the target first adjacent grid. The first position point and the steering angle information constitute the starting part of the predetermined target path. Afterwards, the direction and angle at which the unmanned truck needs to turn after leaving the first position point are determined according to the first position point and the obtained first position steering angle, thereby obtaining the initial segment of the predetermined target path. Steps S3300 to S3500 are re-executed, with the target first adjacent grid as the new search starting point, and the path is continued to be searched to obtain the second position point and the second position steering angle, and finally the complete predetermined target path from the starting grid to the end grid is obtained. The path contains multiple position points and corresponding steering angle information.
[0064] Furthermore, the embodiment of the present application also includes:
[0065] Step S4100: The target operation mode set of the target unmanned truck includes a remote control driving mode, an automatic driving mode and a remote driving mode.
[0066] Specifically, the target operation mode set of the unmanned truck includes three modes: remote control driving mode, automatic driving mode and remote driving mode. In the remote control driving mode, the unmanned truck is directly controlled by humans through control devices such as handles to achieve speed control, steering control and braking control. In this mode, the unmanned truck relies on the operator's control and has high flexibility and environmental adaptability. In the automatic driving mode, the unmanned truck completes the driving task autonomously according to the path information and control instructions provided by the operation control system. The unmanned truck has the ability to track the path and avoid obstacles, and can navigate autonomously in a structured environment efficiently and accurately. The remote driving mode allows the operator at the remote end to send control instructions to the unmanned truck and remotely control its state and movement. In this mode, the operator is farther away from the unmanned truck and relies on wireless communication connection, which has high flexibility. In use, the appropriate mode is selected according to the complexity of the environment and the task requirements, and it can also switch between different modes to achieve higher adaptive control.
[0067] Further, such as Figure 3 As shown, the embodiment of the present application also includes:
[0068] Step S5100: the operation control system is in communication connection with a position sensor, and the position sensor is arranged on the target unmanned truck, and the target real-time position point of the target unmanned truck is obtained through dynamic monitoring by the position sensor;
[0069] Step S5200: Compare the target real-time position point with the predetermined target path to obtain the target real-time steering deviation angle;
[0070] Step S5300: Send the target real-time steering deviation angle to the target unmanned truck for real-time position correction.
[0071] Specifically, position sensors, such as GPS receivers, inertial navigation systems, etc., are installed on the unmanned truck to measure the coordinate position and attitude angle information of the unmanned truck in real time. At the same time, a wireless communication connection is established between the operation control system and the position sensor, and the position sensor sends the measured position information to the operation control system in real time. After the communication module in the operation control system receives the position data sent by the position sensor, it obtains the real-time position point of the unmanned truck, such as the longitude and latitude coordinates, through analysis. Then, the operation control system extracts the target position point of the unmanned truck scheduled for this path segment from the stored predetermined target path data. The target position point and the real-time position point are compared, and the target real-time steering deviation angle of the unmanned truck is calculated based on the previous position point. If the deviation exceeds the predetermined threshold, it is determined that the deviation is too large, and a control instruction needs to be sent for correction.
[0072] Next, the operation control system generates a correction control instruction based on the steering deviation angle, including information such as the steering angle and steering direction; the operation control system sends the correction control instruction to the unmanned truck through a communication connection. After receiving the correction control instruction, the unmanned truck adjusts the steering actuator, such as the motor angle, according to the instruction information, controls the wheel steering, and navigates the unmanned truck to the predetermined target path. After the unmanned truck completes the steering correction, it feeds back the new position and heading information to the operation control system. The operation control system parses the feedback information to determine whether the unmanned truck has returned to the predetermined target path. If it has not returned completely, continue to send correction control instructions until the deviation is reduced to below the threshold. Real-time position correction control of the unmanned truck is realized, the driving safety hazards caused by the deviation of the unmanned truck are minimized, the accuracy of its automatic navigation and transportation efficiency are improved, and the unmanned truck has a strong environmental adaptability.
[0073] Furthermore, the embodiment of the present application also includes:
[0074] Step S5110: the operation control system is in communication connection with a target CCD image sensor, and the target CCD image sensor is arranged in the target operation area, and a target monitoring image time sequence of the target operation area is acquired by the target CCD image sensor, wherein the target monitoring image time sequence includes a plurality of area images;
[0075] Step S5120: extracting the plurality of regional images according to a first image sampling frequency to obtain a first image set;
[0076] Step S5130: extracting and down-sampling the plurality of regional images according to a second image sampling frequency to obtain a second image set, wherein the second image sampling frequency is greater than the first image sampling frequency;
[0077] Step S5140: obtaining an intelligent recognition model based on historical unmanned truck position image data training, and analyzing the first image set and the second image set by the intelligent recognition model to obtain a recognition result, wherein the recognition result refers to a recognition result of the real-time position of the target unmanned truck;
[0078] Step S5150: Obtain the real-time location point of the target based on the recognition result.
[0079] Specifically, multiple target CCD image sensors are set in the target operation area to monitor the environmental image information in the area in real time. The target CCD image sensor is connected to the operation control system through wired or wireless communication, and the collected images are sent to the operation control system in real time. After receiving the image sent by the target CCD image sensor, the operation control system stores it in the image database in chronological order to form a target monitoring image time sequence, which includes multiple area images in the target operation area.
[0080] Then, set the first image sampling frequency, for example, extract an image every 0.5 seconds. According to the first image sampling frequency, extract some images from the target monitoring image sequence in time sequence to form a first image set. Set the second image sampling frequency to be higher than the first frequency, for example, extract an image every 0.2 seconds. According to the second image sampling frequency, extract some images from the target monitoring image sequence in time sequence, perform downsampling on the extracted images, such as reducing the image resolution to reduce the amount of data, and obtain all the downsampled images to form a second image set.
[0081] Next, the historical unmanned truck position image data is collected, which includes multiple historical images collected from the target operation area and the unmanned truck position recognition results corresponding to each image; the historical image data is preprocessed, such as noise filtering, image enhancement, etc.; and then divided into a training set and a test set, the training set is used for model training, and the test set is used to verify the model. The convolutional neural network is selected as the machine learning algorithm, and the supervised learning training is performed with the training set as input. The purpose of the training is to obtain an intelligent recognition model that can identify the position of the target unmanned truck for the input image. Subsequently, the verification set is input into the model, and the recognition results output by the model are compared with the real results corresponding to the verification set. If the recognition accuracy meets the requirements, the model training process ends, otherwise it returns to retrain until the accuracy requirements are met. After the intelligent recognition model meets the requirements, the first image set and the second image set are respectively input into the intelligent recognition model, and the model analyzes and processes the two image sets respectively to obtain two sets of recognition results. The two sets of recognition results are fused to avoid misjudgment of individual results, and the final recognition result of the real-time position of the unmanned truck is obtained. The recognition results are sent to the operation control system as the target real-time location point to track the real-time status and movement position of the unmanned truck.
[0082] Furthermore, the embodiment of the present application also includes:
[0083] Step S5141: the historical unmanned truck position image data includes a plurality of first historical images, a plurality of second historical images, and a plurality of image recognition results;
[0084] Step S5142: Performing supervised learning and verification on the historical unmanned truck location image data to obtain the intelligent recognition model;
[0085] The intelligent recognition model includes a first recognition channel and a second recognition channel, and the input data of the first recognition channel is the first image set, and the input data of the second recognition channel is the second image set.
[0086] Specifically, multiple target CCD image sensors are set in the target operation area to record the operation image information of the unmanned truck for a long time, and the position image of the unmanned truck is collected at two frequencies to obtain the first historical image set and the second historical image set. Each image corresponds to the exact position information of the unmanned truck in the image as the image recognition result. The first historical image set and the second historical image set and the corresponding image recognition results constitute the historical unmanned truck position image data, which are stored in the database of the operation control system. The acquisition frequencies of the first historical image set and the second historical image set are different, but the corresponding images contain the position information of the unmanned truck, but the amount of information is different.
[0087] Then, the historical unmanned truck position image data is used as input, and the convolutional neural network is selected as the supervised learning algorithm to train the intelligent recognition model. The intelligent recognition model contains two parallel recognition channels. The input of the first recognition channel is the first image set, and the input of the second recognition channel is the second image set. The two channels process images of two acquisition frequencies respectively, and each obtains recognition results. The data of the historical first image set and the second image set are preprocessed, such as image enhancement, noise filtering, etc., and divided into training sets and verification sets. The corresponding training sets are input into the first recognition channel and the second recognition channel respectively, and multiple rounds of iterative training are performed to finally obtain two recognition channels that can recognize the position of the unmanned truck for the input image. Subsequently, the corresponding verification set is input into the two channels obtained by training, and the output recognition results are compared with the real results corresponding to the verification set to calculate the recognition accuracy of the channel. If the recognition accuracy does not meet the requirements, return to the retraining channel until the accuracy requirements are met. When both channels reach the required recognition accuracy, the first recognition channel and the second recognition channel are integrated to obtain an intelligent recognition model that meets the requirements.
[0088] Furthermore, the embodiment of the present application also includes:
[0089] Step S6100: The target unmanned truck is provided with a quality sensor, and the target unmanned truck is dynamically monitored for quality by the quality sensor to obtain the target real-time quality;
[0090] Step S6200: Perform integrity monitoring and analysis on the goods on the target unmanned truck according to the target real-time quality.
[0091] Specifically, multiple sets of quality sensors are installed at different locations in the cargo hold of the unmanned truck. The quality sensors use pressure sensors, strain sensors, etc. to regularly detect the real-time quality in the cargo hold of the unmanned truck and send the results to the operation control system. The operation control system receives the real-time quality data sent by the quality sensor, records it as the target real-time quality, and stores it in the database to form a record of the quality change process.
[0092] Subsequently, the operation control system analyzes the stored real-time quality data according to the preset judgment rules to monitor the integrity of the goods. If the real-time quality decreases continuously, it may mean that the goods have fallen off. The operation control system sends a command to the unmanned truck to slow down and stop, and reminds the operator to deal with it if necessary. If the real-time quality suddenly increases or decreases, it may mean that the internal displacement of the goods has affected the load distribution and stability. The operation control system sends a correction command to the unmanned truck, such as adjusting the heading angle to restore the normal distribution of the goods, and then continues to operate.
[0093] By installing quality sensors on unmanned trucks to monitor real-time quality and analyzing the integrity of the transported goods based on quality change information, safety hazards during transportation by unmanned trucks can be effectively reduced, avoiding the scattering or loss of goods.
[0094] In summary, the operation control method of an unmanned truck provided in an embodiment of the present application has the following technical effects: obtaining the target cargo transportation constraints of the target unmanned truck in the target operation area, wherein the target cargo transportation constraints include the cargo transportation starting point and the cargo transportation end point, providing a basis for path planning and operation control; performing multi-dimensional feature collection on the target operation area to obtain the target area information, and drawing the target grid map based on the target area information, providing environmental information for path planning and operation control; analyzing and obtaining a predetermined target path according to the cargo transportation starting point, the cargo transportation end point and the target grid map, wherein the predetermined target path refers to multiple position points with steering angle identifications, providing a control basis for operation control; obtaining the real-time operation mode of the target unmanned truck for selecting an appropriate operation control strategy; in the real-time operation mode, performing operation control on the target unmanned truck in combination with the predetermined target path, realizing high-precision and adaptive operation control of the unmanned truck, and achieving the technical effect of improving the control accuracy and stability of the unmanned truck.
[0095] Embodiment 2
[0096] Based on the same inventive concept as the operation control method of an unmanned truck in the aforementioned embodiment, Figure 3As shown, an embodiment of the present application provides an operation control system of an unmanned truck, the system is communicatively connected with a target unmanned truck, and the system includes:
[0097] A target freight constraint module 11 is used to obtain the target freight constraint of the target unmanned freight vehicle in the target operation area, wherein the target freight constraint includes a freight start point and a freight end point;
[0098] The regional feature acquisition module 12 is used to acquire multi-dimensional features of the target operation area to obtain target area information, and draw a target grid map based on the target area information;
[0099] A predetermined path analysis module 13, configured to analyze and obtain a predetermined target path according to the cargo transportation starting point, the cargo transportation end point and the target grid map, wherein the predetermined target path refers to a plurality of position points with turning angle marks;
[0100] An operation mode acquisition module 14 is used to acquire the real-time operation mode of the target unmanned truck;
[0101] The target operation control module 15 is used to control the operation of the target unmanned truck in combination with the predetermined target path in the real-time operation mode.
[0102] Furthermore, the regional feature acquisition module 12 includes the following execution steps:
[0103] Acquire basic target information of the target operation area, wherein the basic target information includes shape data and size data; establish an initial model of the target operation area according to the shape data and the size data;
[0104] Acquire target layout information of the target operation area, and render the target layout information to the initial model to obtain a target model;
[0105] The target model is grid-divided to obtain a grid-division result, wherein the grid-division result includes a plurality of grids; each of the plurality of grids is judged and analyzed in turn, and the target grid map is obtained according to the analysis result.
[0106] Furthermore, the regional feature acquisition module 12 further includes the following execution steps:
[0107] Extract any one grid from the plurality of grids and record it as a target grid;
[0108] Determine whether the target grid meets the predetermined requirements, if yes, perform a first mark on the target grid, if no, perform a second mark on the target grid;
[0109] The target grid map is constructed based on the first mark or the second mark.
[0110] Furthermore, the predetermined path analysis module 13 includes the following execution steps:
[0111] Matching the first grid of the shipping starting point and the second grid of the shipping destination respectively;
[0112] Reading a predetermined traversal scheme, and obtaining a first adjacent grid set of the first grid based on the predetermined traversal scheme, wherein the first adjacent grid set includes M adjacent grids, and M is an integer greater than or equal to 1;
[0113] Extracting a first adjacent grid from the M adjacent grids, and calculating a first estimated cost from the first adjacent grid to the second grid;
[0114] Taking the first estimated cost as a constraint, selecting the first adjacent grid of the target;
[0115] Taking the first grid as a first position point, combining the first grid and the first adjacent grid of the target to obtain a first position steering angle;
[0116] The predetermined target path is obtained according to the first position point and the first position steering angle.
[0117] Furthermore, the operation mode acquisition module 14 includes the following contents:
[0118] The target operation mode set of the target unmanned truck includes a remote control driving mode, an automatic driving mode and a remote driving mode.
[0119] Furthermore, the target job control module 15 includes the following execution steps:
[0120] The target real-time position point of the target unmanned truck is obtained by dynamic monitoring of the position sensor;
[0121] Comparing the target real-time position point with the predetermined target path to obtain a target real-time steering deviation angle;
[0122] The target real-time steering deviation angle is sent to the target unmanned truck for real-time position correction.
[0123] Furthermore, the target job control module 15 also includes the following execution steps:
[0124] A target monitoring image time sequence of the target operation area is acquired by the target CCD image sensor, wherein the target monitoring image time sequence includes a plurality of area images;
[0125] Extracting the plurality of regional images according to a first image sampling frequency to obtain a first image set;
[0126] Extracting and down-sampling the plurality of regional images according to a second image sampling frequency to obtain a second image set, wherein the second image sampling frequency is greater than the first image sampling frequency;
[0127] An intelligent recognition model is obtained by training based on historical unmanned truck position image data, and the first image set and the second image set are analyzed by the intelligent recognition model to obtain a recognition result, wherein the recognition result refers to a recognition result of the real-time position of the target unmanned truck;
[0128] The real-time location point of the target is obtained based on the recognition result.
[0129] Furthermore, the target job control module 15 also includes the following execution steps:
[0130] The historical unmanned truck location image data includes a plurality of first historical images, a plurality of second historical images, and a plurality of image recognition results;
[0131] Performing supervised learning and verification on the historical unmanned truck location image data to obtain the intelligent recognition model;
[0132] The intelligent recognition model includes a first recognition channel and a second recognition channel, and the input data of the first recognition channel is the first image set, and the input data of the second recognition channel is the second image set.
[0133] Furthermore, the embodiment of the present application also includes a dynamic detection module, which includes the following execution steps:
[0134] Dynamically monitor the quality of the target unmanned truck by using the quality sensor to obtain the target real-time quality;
[0135] The integrity of the goods on the target unmanned truck is monitored and analyzed according to the target real-time quality.
[0136] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.
[0137] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.
Claims
1. An operation control method for an unmanned truck, characterized in that: The operation control method is applied to an operation control system, and the operation control system is in communication connection with a target unmanned transport vehicle, and the operation control method includes: Obtaining target freight constraints of the target unmanned truck in a target operation area, wherein the target freight constraints include a freight start point and a freight end point; Performing multi-dimensional feature collection on the target operation area to obtain target area information, and drawing a target grid map based on the target area information; Analyze and obtain a predetermined target path according to the cargo transportation starting point, the cargo transportation end point and the target grid map, wherein the predetermined target path refers to a plurality of position points with turning angle marks; Obtaining the real-time operation mode of the target unmanned truck; In the real-time operation mode, controlling the operation of the target unmanned truck in combination with the predetermined target path; The step of collecting multi-dimensional features of the target operation area to obtain target area information and drawing a target grid map based on the target area information includes: Acquire basic target information of the target operation area, wherein the basic target information includes shape data and size data; Establishing an initial model of the target operation area according to the shape data and the size data; Acquire target layout information of the target operation area, and render the target layout information to the initial model to obtain a target model; Performing grid division on the target model to obtain a grid division result, wherein the grid division result includes a plurality of grids; Performing judgment and analysis on each of the plurality of grids in turn, and obtaining the target grid map according to the analysis results; The target unmanned transport vehicle is provided with a quality sensor, and the target unmanned transport vehicle is dynamically monitored by the quality sensor to obtain the target real-time quality; The integrity of the goods on the target unmanned truck is monitored and analyzed according to the target real-time quality.
2. The method according to claim 1, characterized in that: The step of sequentially analyzing each grid in the plurality of grids and obtaining the target grid map according to the analysis results includes: Extract any one grid from the plurality of grids and record it as a target grid; Determine whether the target grid meets the predetermined requirements, if yes, perform a first mark on the target grid, if no, perform a second mark on the target grid; The target grid map is constructed based on the first mark or the second mark.
3. The method according to claim 2, characterized in that: The step of analyzing and obtaining a predetermined target path according to the transport starting point, the transport end point and the target grid map includes: Matching the first grid of the shipping starting point and the second grid of the shipping destination respectively; Reading a predetermined traversal scheme, and obtaining a first adjacent grid set of the first grid based on the predetermined traversal scheme, wherein the first adjacent grid set includes M adjacent grids, and M is an integer greater than or equal to 1; Extracting a first adjacent grid from the M adjacent grids, and calculating a first estimated cost from the first adjacent grid to the second grid; Taking the first estimated cost as a constraint, selecting the first adjacent grid of the target; Taking the first grid as a first position point, combining the first grid and the first adjacent grid of the target to obtain a first position steering angle; The predetermined target path is obtained according to the first position point and the first position steering angle.
4. The method according to claim 1, characterized in that: The target operation mode set of the target unmanned truck includes a remote control driving mode, an automatic driving mode and a remote driving mode.
5. The method according to claim 4, characterized in that: The operation control system is in communication with a position sensor, and the position sensor is arranged on the target unmanned truck. In the real-time operation mode, the operation control of the target unmanned truck is performed in combination with the predetermined target path, including: The target real-time position point of the target unmanned truck is obtained by dynamic monitoring of the position sensor; Comparing the target real-time position point with the predetermined target path to obtain a target real-time steering deviation angle; The target real-time steering deviation angle is sent to the target unmanned truck for real-time position correction.
6. The method according to claim 5, characterized in that: The operation control system is in communication with a target CCD image sensor, and the target CCD image sensor is arranged in the target operation area, and the target real-time position point of the target unmanned transport truck is obtained by dynamic monitoring of the position sensor, and further includes: A target monitoring image time sequence of the target operation area is acquired by the target CCD image sensor, wherein the target monitoring image time sequence includes a plurality of area images; Extracting the plurality of regional images according to a first image sampling frequency to obtain a first image set; Extracting and down-sampling the plurality of regional images according to a second image sampling frequency to obtain a second image set, wherein the second image sampling frequency is greater than the first image sampling frequency; An intelligent recognition model is obtained by training based on historical unmanned truck position image data, and the first image set and the second image set are analyzed by the intelligent recognition model to obtain a recognition result, wherein the recognition result refers to a recognition result of the real-time position of the target unmanned truck; The real-time location point of the target is obtained based on the recognition result.
7. The method according to claim 6, characterized in that: The intelligent recognition model obtained by training based on historical unmanned truck location image data includes: The historical unmanned truck location image data includes a plurality of first historical images, a plurality of second historical images, and a plurality of image recognition results; Performing supervised learning and verification on the historical unmanned truck location image data to obtain the intelligent recognition model; The intelligent recognition model includes a first recognition channel and a second recognition channel, and the input data of the first recognition channel is the first image set, and the input data of the second recognition channel is the second image set.
8. An operation control system for an unmanned truck, characterized in that: The method for controlling an operation of an unmanned truck according to any one of claims 1 to 7 is used to implement the operation control system, wherein the operation control system is in communication with a target unmanned truck, and the operation control system comprises: A target freight constraint module, the target freight constraint module is used to obtain the target freight constraint of the target unmanned freight vehicle in the target operation area, wherein the target freight constraint includes a freight start point and a freight end point; A regional feature acquisition module, the regional feature acquisition module is used to perform multi-dimensional feature acquisition on the target operation area to obtain target area information, and draw a target grid map based on the target area information; A predetermined path analysis module, the predetermined path analysis module is used to analyze and obtain a predetermined target path according to the cargo transportation starting point, the cargo transportation end point and the target grid map, wherein the predetermined target path refers to a plurality of position points with steering angle marks; An operation mode acquisition module, the operation mode acquisition module is used to acquire the real-time operation mode of the target unmanned truck; A target operation control module is used to control the operation of the target unmanned truck in combination with the predetermined target path in the real-time operation mode.
