AGV walking state self-adaptive adjustment control method and device
Through lidar and deep learning models, AGV can adaptively adjust the walking state in complex environments, solving the problems of low robustness and cumbersome manual debugging in traditional technologies, and improving safety and adaptability.
Patent Information
- Application Number
- CN202411779617.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-05
- Publication Date
- 2025-01-07
AI Technical Summary
When AGV is walking in complex environments, the traditional map segment settings are low, and the walking state cannot be adjusted adaptively, resulting in cargo offset and collision risks, and manual debugging requires cumbersome and labor-intensive labor.
Point cloud data is collected through lidar, environmental information and risk areas are determined using preset deep learning models, and the walking status of the AGV is adjusted according to the risk level, including reducing the walking speed or changing the route to achieve adaptive adjustment.
Adaptively adjusting the AGV walking state without manual participation improves robustness and safety in complex environments and reduces accident rates and handling costs.
Smart Images
Figure CN119261964A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of guided vehicle control, and in particular to a control method and device, apparatus, equipment and storage medium for adaptively adjusting the walking state of an AGV. Background Art
[0002] During operation, AGVs (Automated Guided Vehicles) often encounter complex environments such as rough roads, potholes, water accumulation, oil stains, and narrow lanes. At the same time, due to the diversity of the goods being transported, when the sliding friction of the goods becomes smaller, the goods will shift, resulting in the risk of collision.
[0003] At present, in the field of automated warehousing and logistics, AGV unmanned forklifts have been widely used in cargo handling and warehouse operations. However, different industries and factories have their own unique complex environments. When facing complex challenges, AGV can only rely on on-site debugging personnel to set different map segment parameters according to different on-site environments. However, this parameter configuration is very cumbersome and cannot cope with emergencies. At the same time, the use of traditional map segments to set AGV walking routes and speed schemes has extremely low robustness and consumes a lot of manpower and material resources. When the environment of the configured AGV walking route changes, the AGV will still walk according to the set route and speed without adjustment.
[0004] In summary, how to adaptively adjust the walking state of AGV without human intervention is a technical problem that needs to be solved urgently. Summary of the invention
[0005] In view of this, the purpose of the present invention is to provide a control method and device for adaptively adjusting the walking state of an AGV, which can adaptively adjust the walking state of the AGV without human intervention. The specific scheme is as follows:
[0006] In a first aspect, the present application provides a control method for adaptively adjusting the walking state of an AGV, comprising:
[0007] Determine the corresponding target range according to the current position of the automatic guided vehicle, and obtain the point cloud data within the target range collected by the laser radar;
[0008] Determine the data type of the point cloud data by using a preset deep learning model, and determine environmental information within the target range according to the data type of the point cloud data, so as to construct a risk area corresponding to the automated guided vehicle based on the environmental information;
[0009] The current risk level of the automatic guided vehicle is determined based on the environmental information, and the walking state of the automatic guided vehicle is adjusted based on the risk level and the risk area.
[0010] Optionally, determining the data type of the point cloud data by using a preset deep learning model, and determining the environmental information within the target range according to the data type of the point cloud data, includes:
[0011] If the point cloud data is determined to be ground point cloud data through the preset deep learning model, a plane normal vector of the ground point cloud data is calculated to determine the inclination angle of the ground relative to the horizontal plane and the thickness of the ground point cloud according to the plane normal vector;
[0012] Accordingly, determining the current risk level of the automatic guided vehicle based on the environmental information includes:
[0013] If the tilt angle is greater than a preset tilt angle threshold, determining that the current risk level of the automated guided vehicle is a level one risk level;
[0014] If the thickness of the ground point cloud is greater than a first preset thickness threshold and less than a second preset thickness threshold, determining that the current risk level of the automated guided vehicle is a secondary risk level;
[0015] If the thickness of the ground point cloud is not less than the second preset thickness threshold, it is determined that the current risk level of the automatic guided vehicle is a third risk level.
[0016] Optionally, determining the data type of the point cloud data by using a preset deep learning model, and determining the environmental information within the target range according to the data type of the point cloud data, includes:
[0017] If the point cloud data is determined to be ground point cloud data through the preset deep learning model, and it is determined that there is a preset sparse area in the ground point cloud data whose data density is less than a preset density, then it is determined that there is a first target area in the target range;
[0018] If the point cloud data is determined to be ground point cloud data through the preset deep learning model, and it is determined that a preset hole area exists in the ground point cloud data, then it is determined that a second target area exists within the target range.
[0019] Optionally, determining the current risk level of the automated guided vehicle based on the environmental information includes:
[0020] When the first target area exists within the target range, determining a difference between a minimum value of a regional point cloud in the preset sparse area corresponding to the first target area and a value of a ground point cloud plane;
[0021] If the difference is less than a preset difference threshold, determining that the risk level of the automated guided vehicle in the risk area is a secondary risk level;
[0022] If the difference is not less than a preset difference threshold, determining that the risk level of the automated guided vehicle in the risk area is a third risk level;
[0023] When the second target area exists within the target range, the risk level of the automatic guided vehicle in the risk area is directly determined to be a third risk level.
[0024] Optionally, determining the data type of the point cloud data by using a preset deep learning model, and determining the environmental information within the target range according to the data type of the point cloud data, includes:
[0025] If the point cloud data is determined to be wall point cloud data through the preset deep learning model, then determine whether there is a lane to be passed within the target range, and if so, determine the distance value between the wall point cloud data, and determine the lane width of the lane to be passed according to the distance value;
[0026] Accordingly, determining the current risk level of the automatic guided vehicle based on the environmental information includes:
[0027] Determining a difference between the lane width and a maximum width value of the automated guided vehicle;
[0028] If the difference is not less than the first preset distance threshold and less than the second preset distance threshold, determining that the risk level of the automated guided vehicle in the risk area is a first-level risk level;
[0029] If the difference is not less than the third preset distance threshold and less than the first preset distance threshold, determining that the risk level of the automated guided vehicle in the risk area is a secondary risk level;
[0030] If the distance is less than a third preset distance threshold, it is determined that the risk level of the automatic guided vehicle in the risk area is a third risk level.
[0031] Optionally, determining the data type of the point cloud data by using a preset deep learning model, and determining the environmental information within the target range according to the data type of the point cloud data, includes:
[0032] If the point cloud data is determined to be human point cloud data by the preset deep learning model, the movement speed, the direction of travel and the number of human bodies within the target range are determined by the human point cloud data;
[0033] Accordingly, determining the current risk level of the automatic guided vehicle based on the environmental information includes:
[0034] Determining an encounter time between the human body and the automated guided vehicle based on the movement speed and the travel direction, and determining a time difference between the encounter time and a current time;
[0035] If the number of human bodies in the current risk area is less than a preset number threshold, determining that the current risk level of the automated guided vehicle is a level one risk level;
[0036] If the number of human bodies in the current risk area is not less than a preset number threshold, determining that the current risk level of the automated guided vehicle is a level 2 risk level;
[0037] If the time difference is less than the first preset time threshold, it is determined that the current risk level of the automatic guided vehicle is a third risk level.
[0038] Optionally, adjusting the walking state of the automatic guided vehicle based on the risk level and the risk area includes:
[0039] If the current risk level of the automated guided vehicle is level one, reducing the walking speed of the automated guided vehicle in the risk area according to a first preset ratio;
[0040] If the current risk level of the automated guided vehicle is level 2, reducing the walking speed of the automated guided vehicle in the risk area according to a second preset ratio;
[0041] If the current risk level of the automated guided vehicle is level three, controlling the automated guided vehicle to stop moving, and updating the moving route of the automated guided vehicle in the risk area;
[0042] Correspondingly, after adjusting the walking state of the automatic guided vehicle based on the risk level and the risk area, the method further includes:
[0043] If it is determined that the automated guided vehicle passes through the risk area, the current risk level of the automated guided vehicle is canceled and the process jumps to the step of determining the corresponding target range according to the current position of the automated guided vehicle and acquiring the point cloud data within the target range collected by the laser radar.
[0044] In a second aspect, the present application provides a control device for adaptively adjusting the walking state of an AGV, comprising:
[0045] A data collection module, used to determine the corresponding target range according to the current position of the automatic guided vehicle, and obtain point cloud data within the target range collected by the laser radar;
[0046] A data processing module, used to determine the data type of the point cloud data through a preset deep learning model, and determine the environmental information within the target range according to the data type of the point cloud data, so as to construct a risk area corresponding to the automatic guided vehicle based on the environmental information;
[0047] The guided vehicle control module is used to determine the current risk level of the automatic guided vehicle based on the environmental information, and adjust the walking state of the automatic guided vehicle based on the risk level and the risk area.
[0048] In this application, firstly, the corresponding target range is determined according to the current position of the automatic guided vehicle, and the point cloud data within the target range collected by the laser radar is obtained; then the point cloud data is processed by a preset deep learning model to determine the environmental information corresponding to the point cloud data; finally, the current risk level of the automatic guided vehicle is determined based on the environmental information, and the risk area corresponding to the automatic guided vehicle is constructed based on the environmental information, and the walking state of the automatic guided vehicle is adjusted based on the risk level and the risk area. As can be seen from the above, the present application can obtain the point cloud data within the target range of the automatic guided vehicle collected by the laser radar, and process the point cloud data to determine the risk level and corresponding risk area of the automatic guided vehicle, and then adjust the walking state of the automatic guided vehicle. In this way, the environmental information around the AGV can be obtained by the laser radar without human intervention to determine the risk level of the AGV and construct the walking risk area of the AGV, so as to adaptively adjust the walking state of the AGV, so that the AGV can pass through the risk area at different safe speeds, or change the route. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0050] Figure 1 A flow chart of a control method for adaptively adjusting the walking state of an AGV provided in this application;
[0051] Figure 2 A flow chart of a specific control method for adaptive adjustment of AGV walking state provided in this application;
[0052] Figure 3 A system framework diagram of a control method for adaptively adjusting the walking state of an AGV provided in this application;
[0053] Figure 4A schematic diagram of the structure of a control device for adaptively adjusting the walking state of an AGV provided in this application;
[0054] Figure 5 A structural diagram of an electronic device provided for this application. DETAILED DESCRIPTION
[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0056] At present, in the field of automated warehousing and logistics, AGV unmanned forklifts have been widely used in cargo handling and warehouse operations. However, different industries and factories have their own unique complex environments. When facing complex challenges, AGV can only rely on on-site debugging personnel to set different map segment parameters according to different on-site environments. However, this parameter configuration is very cumbersome and cannot cope with emergencies. At the same time, the use of traditional map segments to set the AGV walking route and speed scheme has extremely low robustness and consumes a lot of manpower and material resources. When the environment changes on the configured AGV walking route, the AGV will still walk according to the set route and speed and will not be adjusted. To this end, the present application provides a solution for adaptive adjustment of the walking state of the AGV, which can adaptively adjust the walking state of the AGV without human intervention.
[0057] See also Figure 1 As shown, the embodiment of the present application discloses a control method for adaptively adjusting the walking state of an AGV, which may include:
[0058] Step S11, determine the corresponding target range according to the current position of the automatic guided vehicle, and obtain point cloud data within the target range collected by the laser radar.
[0059] In this embodiment, sensors and navigation technology can be used to monitor the current position and driving status of the automatic guided vehicle in real time, and then the obstacle detection area within a certain range around the automatic guided vehicle, i.e., the target range, is determined based on the current position and speed of the automatic guided vehicle in combination with the distance sensor.
[0060] In a specific implementation, the automated guided vehicle may be equipped with a laser radar as a sensor. The laser radar device may emit laser pulses to a target range. When the laser pulses encounter an object, part of the light wave will be reflected back and captured by the laser radar receiver. The laser radar receiver converts the received light signal into an electrical signal to subsequently generate three-dimensional point cloud data within the target range.
[0061] In another specific implementation, the laser radar can be installed at a fixed position on the ground to ensure that the laser radar can scan the environmental information within the target range of the automatic guided vehicle. At the same time, the configuration parameters of the laser radar, such as scanning range, resolution, scanning frequency, etc., can be adjusted according to the application requirements and working environment of the laser radar.
[0062] It is understandable that the point cloud data collected by the LiDAR can provide the AGV with detailed environmental information, including information on the ground, walls, water, obstacles, other vehicles, and pedestrians. Based on the point cloud data, the AGV can generate the optimal driving path to ensure that the AGV can reach its destination safely. At the same time, during driving, the AGV can use point cloud data to monitor changes in the surrounding environment in real time, such as new obstacles or road changes. Based on this environmental information, the AGV can adjust its driving strategy in a timely manner and navigate the route to safely avoid obstacles.
[0063] Step S12: determine the data type of the point cloud data through a preset deep learning model, and determine the environmental information within the target range according to the data type of the point cloud data, so as to construct a risk area corresponding to the automatic guided vehicle based on the environmental information.
[0064] In this embodiment, after the automatic guided vehicle obtains the point cloud data within the target range collected by the laser radar, the point cloud data can be sent to the edge computing module, and the edge computing module can identify the data type of the point cloud data through the deep learning model and perform classification processing. In order to train the deep learning model, a point cloud data set containing different data types can be collected in advance, and then each point cloud data in the point cloud data set can be pre-processed, including denoising, filtering, standardization, etc., to improve the accuracy of model recognition. Deep learning models such as PointNet, PointNet++, PointCNN (CNN, Convolutional Neural Networks, Convolutional Neural Networks) can be used to extract useful features from point cloud data, such as local geometric structure, global features, color information, etc. of point cloud data. Then the extracted features and corresponding labels are input into the deep learning model for model training, so that the deep learning model determines the mapping relationship between point cloud features and data types. Finally, after the model training is completed, the deep learning model can be verified using a test data set to determine whether the deep learning model can output the data type label of each point or the entire point cloud based on the input point cloud data.
[0065] In a first specific implementation, the data type of the point cloud data is determined by a preset deep learning model, and the environmental information within the target range is determined according to the data type of the point cloud data, which may include: if the point cloud data is determined to be ground point cloud data by the preset deep learning model, the plane normal vector of the ground point cloud data is calculated to determine the inclination angle of the ground relative to the horizontal plane and the thickness of the ground point cloud according to the plane normal vector. Specifically, the point cloud data within the target range collected by the laser radar is input into the preset deep learning model of the edge computing module of the AGV. If the preset deep learning model identifies that the data type of the point cloud data is ground point cloud data, the ground point cloud data can be plane-fitted by the least squares method to determine the plane equation corresponding to the ground point cloud data to extract the plane corresponding to the ground point cloud. According to the plane equation of the ground, the plane normal vector of the ground point cloud data can be directly calculated, and then the inclination angle of the ground relative to the laser radar can be obtained. After adding the laser radar installation angle data, the inclination angle of the ground relative to the horizontal plane is calculated. At the same time, the ground point cloud data is traversed to find the maximum and minimum values of all points in the vertical direction to obtain the thickness of the ground point cloud.
[0066] In a second specific implementation, the data type of the point cloud data is determined by a preset deep learning model, and the environmental information within the target range is determined according to the data type of the point cloud data, which may include: if the point cloud data is determined to be ground point cloud data by the preset deep learning model, and it is determined that there is a preset sparse area with a data density less than a preset density in the ground point cloud data, then it is determined that there is a first target area within the target range. Specifically, the point cloud data within the target range collected by the laser radar is input into the preset deep learning model of the edge computing module of the AGV, and the preset deep learning model processes the input point cloud data, outputs the density value of each grid or each point, and marks the area with a data density less than the preset density as a sparse area according to the preset density. If it is determined that there is a preset sparse area with a data density less than a preset density in the ground point cloud data, then it is determined that there is a first target area within the target range, that is, water accumulation occurs in the first target area. Since the water accumulation area reflects the laser radar beam weakly, the point cloud data in the area with water accumulation is relatively sparse.
[0067] In a third specific implementation, the data type of the point cloud data is determined by a preset deep learning model, and the environmental information within the target range is determined based on the data type of the point cloud data, which may include: if the point cloud data is determined to be ground point cloud data by the preset deep learning model, and it is determined that there is a preset hole area in the ground point cloud data, then it is determined that there is a second target area within the target range. Specifically, the point cloud data within the target range collected by the laser radar is input into the preset deep learning model of the edge computing module of the AGV, and the preset deep learning model determines that there is a second target area within the target range where point cloud data is missing. It is understandable that when a preset hole area appears in the ground point cloud data, it indicates that oil stains appear in the area. Because the oil stains absorb part of the laser radar's transmitted laser and reflect most of the transmitted laser mirror, the transmitted laser cannot be returned and absorbed by the laser radar.
[0068] In this embodiment, the AGV obtains the point cloud data within the target range collected by the laser radar and sends the point cloud data to the edge computing module. The edge computing module of the AGV determines that there are water accumulation areas or oil stains in the ground point cloud data through a preset deep learning model, and then sets the water accumulation area or oil stain area as the risk area corresponding to the automatic guided vehicle. After that, the edge computing module transmits the coordinates of the risk area to the vehicle control module through the communication module, so that the vehicle control module can adjust the walking state of the vehicle.
[0069] Step S13: determining a current risk level of the automatic guided vehicle based on the environmental information, and adjusting a walking state of the automatic guided vehicle based on the risk level and the risk area.
[0070] In a first specific implementation, if the point cloud data is determined to be ground point cloud data through a preset deep learning model, the above-mentioned determination of the current risk level of the automatic guided vehicle based on environmental information may include: if the inclination angle is greater than a preset inclination angle threshold, the risk level of the current automatic guided vehicle is determined to be a first-level risk level; if the thickness of the ground point cloud is greater than the first preset thickness threshold and less than the second preset thickness threshold, the risk level of the current automatic guided vehicle is determined to be a second-level risk level; if the thickness of the ground point cloud is not less than the second preset thickness threshold, the risk level of the current automatic guided vehicle is determined to be a third-level risk level. Specifically, if the inclination angle of the ground relative to the horizontal plane is greater than 2°, the edge computing module determines that the risk level of the current AGV is a level one risk level, and sends the corresponding level one risk warning information to the vehicle control module through the communication module; if the thickness of the ground point cloud is greater than 20mm and less than 50mm, it means that the ground in this area is relatively rough and uneven, then the edge computing module determines that the risk level of the current AGV is a level two risk level, and sends the corresponding level two risk warning information to the vehicle control module through the communication module; if the thickness of the ground point cloud is greater than or equal to 50mm, it means that there are potholes on the ground in this area, then the edge computing module determines that the risk level of the current AGV is a level three risk level, and sends the corresponding level three risk warning information to the vehicle control module through the communication module.
[0071] In a second specific implementation, if the point cloud data is determined to be ground point cloud data through a preset deep learning model, and it is determined that there is a first target area within the target range that is a preset sparse area, then the above-mentioned determination of the current risk level of the automatic guided vehicle based on environmental information may include: when there is a first target area within the target range, determining the difference between the minimum value of the regional point cloud within the preset sparse area corresponding to the first target area and the value of the ground point cloud plane; if the above difference is less than the preset difference threshold, determining that the risk level of the automatic guided vehicle in the risk area is a second-level risk level; if the difference is not less than the preset difference threshold, determining that the risk level of the automatic guided vehicle in the risk area is a third-level risk level. Specifically, if the difference between the minimum value of the regional point cloud in the preset sparse area and the value of the ground point cloud plane is less than 50mm, it means that the water in the area is not deep, and the edge computing module determines that the risk level of the current AGV is the second-level risk level, and sends the corresponding second-level risk warning information to the vehicle control module through the communication module; if the difference between the minimum value of the regional point cloud in the preset sparse area and the value of the ground point cloud plane is greater than or equal to 50mm, it means that the water in the area is deep and the AGV cannot wade through, and the edge computing module determines that the risk level of the current AGV is the third-level risk level, and sends the corresponding third-level risk warning information to the vehicle control module through the communication module.
[0072] In a third specific implementation, if the point cloud data is determined to be ground point cloud data through a preset deep learning model, and it is determined that there is a second target area that is a preset hole area within the target range, then the above-mentioned determination of the current risk level of the automatic guided vehicle based on environmental information may include: when there is a second target area within the target range, directly determining the risk level of the automatic guided vehicle in the risk area as a third-level risk level. Specifically, if it is determined that there is a preset hole area in the ground point cloud data, the edge computing module determines that the risk level of the current AGV is a third-level risk level, and sends the corresponding third-level risk warning information to the vehicle control module through the communication module.
[0073] In this embodiment, the above-mentioned adjustment of the walking state of the automatic guided vehicle based on the risk level and the risk area may include: if the current risk level of the automatic guided vehicle is the first risk level, then the walking speed of the automatic guided vehicle in the risk area is reduced according to the first preset ratio; if the current risk level of the automatic guided vehicle is the second risk level, then the walking speed of the automatic guided vehicle in the risk area is reduced according to the second preset ratio; if the current risk level of the automatic guided vehicle is the third risk level, then the automatic guided vehicle is controlled to stop walking, and the walking route of the automatic guided vehicle in the risk area is updated. In a specific embodiment, if the risk level of the automatic guided vehicle is determined to be the first risk level, the vehicle control module of the AGV will control the AGV to slow down by 10% to pass through the corresponding risk area; if the risk level of the automatic guided vehicle is determined to be the second risk level, then the vehicle control module of the AGV will control the AGV to slow down by 50% to pass through the corresponding risk area; if the risk level of the automatic guided vehicle is determined to be the third risk level, then the vehicle control module of the AGV will control the AGV to stop walking, and update the walking route of the AGV in the corresponding risk area.
[0074] As can be seen from the above, in this embodiment, the corresponding target range is first determined according to the current position of the automatic guided vehicle, and the point cloud data within the target range collected by the laser radar is obtained; then the point cloud data is processed by a preset deep learning model to determine the environmental information corresponding to the point cloud data; finally, the current risk level of the automatic guided vehicle is determined based on the environmental information, and the risk area corresponding to the automatic guided vehicle is constructed based on the environmental information, and the walking state of the automatic guided vehicle is adjusted based on the risk level and the risk area. As can be seen from the above, this embodiment can obtain the point cloud data within the target range of the automatic guided vehicle collected by the laser radar, and determine the ground information within the target range of the automatic guided vehicle based on the point cloud data to determine the risk level of the automatic guided vehicle and the corresponding risk area, and then adjust the walking state of the automatic guided vehicle. In this way, the ground information around the AGV can be obtained by the laser radar without human intervention to determine the risk level of the AGV and construct the walking risk area of the AGV, so as to adaptively adjust the walking state of the AGV, so that the AGV can pass through the risk area at different safe speeds, or change the route.
[0075] Based on the previous embodiment, it can be known that the present application can obtain the ground information around the AGV through the laser radar, determine the risk level of the automatic guided vehicle based on the ground information and construct the AGV walking risk area to adaptively adjust the walking state of the AGV. Next, this embodiment will process other environmental information around the AGV obtained by the laser radar to adaptively adjust the walking state of the AGV. Figure 2 As shown, the embodiment of the present application further discloses a specific control method for adaptively adjusting the walking state of an AGV, which may include:
[0076] Step S21, determine the corresponding target range according to the current position of the automatic guided vehicle, and obtain the point cloud data within the target range collected by the laser radar.
[0077] Step S22: determine the data type of the point cloud data through a preset deep learning model, and determine the environmental information within the target range according to the data type of the point cloud data, so as to construct a risk area corresponding to the automatic guided vehicle based on the environmental information.
[0078] In a specific embodiment, the above-mentioned determination of the data type of the point cloud data by a preset deep learning model, and determination of the environmental information within the target range according to the data type of the point cloud data, may include: if the point cloud data is determined to be wall point cloud data by the preset deep learning model, then determining whether there is a lane to be passed within the target range, and if so, determining the distance value between the wall point cloud data, and determining the lane width of the lane to be passed according to the distance value. Specifically, the point cloud data within the target range collected by the laser radar is input into the preset deep learning model of the edge computing module of the AGV. If the preset deep learning model identifies that the data type of the point cloud data is wall point cloud data, and determines that there is a lane to be passed within the target range, then the distance value between the wall point cloud data can be determined by a preset measurement method to determine the lane width of the lane to be passed. Among them, the preset measurement method includes but is not limited to: Euclidean distance, nearest neighbor distance, Chebyshev distance, etc.
[0079] In another specific implementation, the above-mentioned determination of the data type of the point cloud data by a preset deep learning model, and determination of the environmental information within the target range according to the data type of the point cloud data, may include: if the point cloud data is determined to be human point cloud data by the preset deep learning model, the movement speed, travel direction and number of human bodies within the target range are determined by the human point cloud data. Specifically, the point cloud data within the target range collected by the laser radar is input into the preset deep learning model of the edge computing module of the AGV. If the preset deep learning model recognizes that the data type of the point cloud data is human point cloud data, the distance value and the number of human bodies of the human point cloud data are calculated, and the movement speed and travel direction of the human body are calculated according to the change in the distance value of the human point cloud data.
[0080] Step S23: determining a current risk level of the automatic guided vehicle based on the environmental information, and adjusting a walking state of the automatic guided vehicle based on the risk level and the risk area.
[0081] In a specific embodiment, if the point cloud data is determined to be wall point cloud data through the preset deep learning model, and it is determined that there is a lane to be passed within the target range, the distance value between the wall point cloud data is determined, and the lane width of the lane to be passed is determined according to the distance value. The above-mentioned determination of the current risk level of the automatic guided vehicle based on the environmental information may include: determining the difference between the lane width and the maximum width value of the automatic guided vehicle; if the difference is not less than the first preset distance threshold and less than the second preset distance threshold, determining that the risk level of the automatic guided vehicle in the risk area is a first-level risk level; if the difference is not less than the third preset distance threshold and less than the first preset distance threshold, determining that the risk level of the automatic guided vehicle in the risk area is a second-level risk level; if the distance is less than the third preset distance threshold, determining that the risk level of the automatic guided vehicle in the risk area is a third-level risk level. Specifically, the edge computing module of the AGV calculates the difference between the lane width and the maximum width of the automated guided vehicle. If the difference is greater than or equal to 500mm and less than 1000mm, it means that the AGV is about to pass through a narrow lane with a clearance greater than 500mm. The edge computing module determines that the current risk level of the AGV is a level 1 risk level, and sends the corresponding level 1 risk warning information to the vehicle control module through the communication module. After that, the vehicle control module of the AGV will control the AGV to slow down by 10% to pass through the corresponding risk area. If the difference is greater than or equal to 200mm and less than 500mm, it means that the AGV is about to pass through an extremely narrow lane with a clearance greater than 200mm. If the lane is narrow, the edge computing module determines that the current AGV’s risk level is a level 2 risk level, and sends the corresponding level 2 risk warning information to the vehicle control module through the communication module. After that, the vehicle control module of the AGV will control the AGV to slow down by 50% to pass through the corresponding risk area. If the difference is less than 200mm, it means that the narrow lane is not suitable for AGV to pass. Then, the edge computing module determines that the current AGV’s risk level is a level 3 risk level, and sends the corresponding level 3 risk warning information to the vehicle control module through the communication module. After that, the vehicle control module of the AGV will control the AGV to stop walking and update the walking route of the AGV in the corresponding risk area.
[0082] In another specific embodiment, if the point cloud data is determined to be human point cloud data through the preset deep learning model, the movement speed, travel direction and number of human bodies within the target range are determined through the human point cloud data. The above-mentioned determination of the current risk level of the automatic guided vehicle based on the environmental information may include: determining the encounter time between the human body and the automatic guided vehicle based on the movement speed and the travel direction, and determining the time difference between the encounter time and the current time; if the number of human bodies in the current risk area is less than a preset number threshold, the current risk level of the automatic guided vehicle is determined to be a first-level risk level; if the number of human bodies in the current risk area is not less than the preset number threshold, the current risk level of the automatic guided vehicle is determined to be a second-level risk level; if the time difference is less than the first preset time threshold, the current risk level of the automatic guided vehicle is determined to be a third-level risk level. Specifically, the edge computing module of the AGV determines the encounter time between the human body and the automatic guided vehicle according to the change in the distance value of the human body point cloud data in real time, and calculates the time difference between the encounter time and the current time. If the number of human bodies detected within a range of 10 meters around the AGV is less than 3, the edge computing module determines that the current risk level of the AGV is a level 1 risk level, and sends the corresponding level 1 risk warning information to the vehicle control module through the communication module. After that, the vehicle control module of the AGV will control the AGV to slow down by 10% to pass through the corresponding risk area; if the number of human bodies detected within a range of 10 meters around the AGV is greater than or equal to 3, the edge computing module determines The current risk level of the AGV is the second-level risk level, and the corresponding second-level risk warning information is sent to the vehicle control module through the communication module. After that, the vehicle control module of the AGV will control the AGV to slow down by 10% to pass through the corresponding risk area; if it is judged that the human body will come into contact with the AGV within 5 seconds, the edge computing module determines that the current risk level of the AGV is the third-level risk level, and sends the corresponding third-level risk warning information to the vehicle control module through the communication module. After that, the vehicle control module of the AGV will control the AGV to stop walking and wait for the human body to move away. When the human body is away for more than 120 seconds, the vehicle control module will sound the horn to warn.
[0083] Step S24: If it is determined that the automated guided vehicle passes through the risk area, the current risk level of the automated guided vehicle is canceled and the process jumps to the step of determining the corresponding target range according to the current position of the automated guided vehicle, and obtaining the point cloud data within the target range collected by the laser radar.
[0084] In this embodiment, after adjusting the walking state of the AGV, real-time monitoring is performed to determine whether the AGV passes through the risk area. If the AGV passes through the risk area, the laser radar can continue to collect point cloud data within the target range corresponding to the current position of the AGV to determine the environmental information around the AGV. If there is still a risk area, the walking state of the AGV is adaptively adjusted.
[0085] See also Figure 3 As shown in the system framework diagram of a control method for adaptive adjustment of AGV walking state, the process of controlling the guide vehicle can be: the AGV collects various types of point cloud data such as the ground, water, wall, obstacles, and human body in the environment through the laser radar, and provides the point cloud data to the edge computing module. After the edge computing module identifies the data type of the point cloud data through the deep learning model, it classifies and processes the point cloud data to obtain the corresponding environmental information, thereby establishing the risk level of different AGVs and constructing the risk area of the corresponding range. After that, the edge computing module transmits the risk level and the coordinates of the risk area to the vehicle control module through the communication module, so that the vehicle control module can adjust the walking state of the AGV.
[0086] For a more specific processing procedure of the above step S21, reference may be made to the corresponding contents disclosed in the above embodiments, which will not be described in detail here.
[0087] As can be seen from the above, this embodiment can obtain the point cloud data within the target range of the automatic guided vehicle collected by the laser radar, and determine the wall information and human body information within the target range of the automatic guided vehicle based on the point cloud data, so as to determine the risk level and corresponding risk area of the automatic guided vehicle, and then adjust the walking state of the automatic guided vehicle. In this way, the present application can detect the surrounding environment in real time when the AGV is running, and use the edge computing module to classify the risk level of the risk area of the AGV in different environment types according to the surrounding environment information, so as to adaptively adjust the walking state of the AGV. In this way, by collecting and analyzing the data of the surrounding environment of the AGV, risk areas of AGVs with different risk levels can be constructed to adaptively adjust the walking state of the AGV. There is no need to spend manpower to configure complicated map parameters, which saves corresponding manpower and material resources, and effectively improves the adaptability and robustness of the AGV to complex environments, ensures the walking safety of the AGV during operation, and reduces the accident rate and accident handling costs.
[0088] Accordingly, see Figure 4 As shown, the embodiment of the present application also provides a control device for adaptively adjusting the walking state of an AGV, which may include:
[0089] The data collection module 11 is used to determine the corresponding target range according to the current position of the automatic guided vehicle, and obtain the point cloud data within the target range collected by the laser radar;
[0090] A data processing module 12 is used to determine the data type of the point cloud data through a preset deep learning model, and determine the environmental information within the target range according to the data type of the point cloud data, so as to construct a risk area corresponding to the automatic guided vehicle based on the environmental information;
[0091] The guided vehicle control module 13 is used to determine the current risk level of the automatic guided vehicle based on the environmental information, and adjust the walking state of the automatic guided vehicle based on the risk level and the risk area.
[0092] As can be seen from the above, in this application, the corresponding target range is first determined according to the current position of the automatic guided vehicle, and the point cloud data within the target range collected by the laser radar is obtained; then the point cloud data is processed by a preset deep learning model to determine the environmental information corresponding to the point cloud data; finally, the current risk level of the automatic guided vehicle is determined based on the environmental information, and the risk area corresponding to the automatic guided vehicle is constructed based on the environmental information, and the walking state of the automatic guided vehicle is adjusted based on the risk level and risk area. As can be seen from the above, the present application can obtain the point cloud data within the target range of the automatic guided vehicle collected by the laser radar, and process the point cloud data to determine the risk level and corresponding risk area of the automatic guided vehicle, and then adjust the walking state of the automatic guided vehicle. In this way, the environmental information around the AGV can be obtained by the laser radar without human intervention to determine the risk level of the AGV and construct the walking risk area of the AGV, so as to adaptively adjust the walking state of the AGV, so that the AGV can pass through the risk area at different safe speeds, or change the route.
[0093] In some specific implementations, the data processing module 12 includes:
[0094] An angle and thickness determination unit, configured to calculate a plane normal vector of the ground point cloud data if the point cloud data is determined to be ground point cloud data by the preset deep learning model, so as to determine an inclination angle of the ground relative to a horizontal plane and a thickness of the ground point cloud according to the plane normal vector;
[0095] Accordingly, the guided vehicle control module 13 includes:
[0096] an angle comparison unit, configured to determine that the current risk level of the automated guided vehicle is a first-level risk level if the tilt angle is greater than a preset tilt angle threshold;
[0097] A first thickness comparison unit, configured to determine that the current risk level of the automated guided vehicle is a secondary risk level if the thickness of the ground point cloud is greater than a first preset thickness threshold and less than a second preset thickness threshold;
[0098] The second thickness comparison unit is used to determine that the current risk level of the automatic guided vehicle is a third risk level if the thickness of the ground point cloud is not less than a second preset thickness threshold.
[0099] In some specific implementations, the data processing module 12 includes:
[0100] A first target area determination unit, configured to determine that a first target area exists within the target range if the point cloud data is determined to be ground point cloud data through the preset deep learning model and a preset sparse area having a data density less than a preset density exists in the ground point cloud data;
[0101] The second target area determination unit is used to determine whether there is a second target area within the target range if the point cloud data is determined to be ground point cloud data through the preset deep learning model and a preset hole area is determined to exist in the ground point cloud data.
[0102] In some specific implementations, the guided vehicle control module 13 includes:
[0103] A difference determination unit, configured to determine, when the first target area exists within the target range, a difference between a minimum value of a regional point cloud in the preset sparse area corresponding to the first target area and a value of a ground point cloud plane;
[0104] A first difference comparison unit is used to determine that the risk level of the automated guided vehicle in the risk area is a secondary risk level if the difference is less than a preset difference threshold;
[0105] a second difference comparison unit, configured to determine that the risk level of the automated guided vehicle in the risk area is a third risk level if the difference is not less than a preset difference threshold;
[0106] The second target area judgment unit is used to directly determine that the risk level of the automatic guided vehicle in the risk area is a third risk level when the second target area exists in the target range.
[0107] In some specific implementations, the data processing module 12 includes:
[0108] A lane width determination unit, for determining whether there is a lane to be passed within the target range if the point cloud data is determined to be wall point cloud data through the preset deep learning model, and if so, determining the distance value between the wall point cloud data, and determining the lane width of the lane to be passed according to the distance value;
[0109] Correspondingly, the guided vehicle control module 13 includes:
[0110] a width difference determination unit, used to determine the difference between the lane width and the maximum width value of the automated guided vehicle;
[0111] a first width difference determination unit, configured to determine that the risk level of the automated guided vehicle in the risk area is a first risk level if the difference is not less than a first preset distance threshold and less than a second preset distance threshold;
[0112] a second width difference determination unit, configured to determine that the risk level of the automated guided vehicle in the risk area is a second risk level if the difference is not less than a third preset distance threshold and less than the first preset distance threshold;
[0113] The third width difference determination unit is configured to determine that the risk level of the automatic guided vehicle in the risk area is a third risk level if the distance is less than a third preset distance threshold.
[0114] In some specific implementations, the data processing module 12 includes:
[0115] A human body information determination unit, configured to determine the movement speed, travel direction and number of human bodies within the target range through the human body point cloud data if the point cloud data is determined to be human body point cloud data through the preset deep learning model;
[0116] Accordingly, the guided vehicle control module 13 includes:
[0117] a time difference determination unit, configured to determine an encounter time between the human body and the automated guided vehicle based on the movement speed and the travel direction, and to determine a time difference between the encounter time and a current time;
[0118] A first number comparison unit is used to determine that the current risk level of the automated guided vehicle is a first-level risk level if the number of human bodies in the current risk area is less than a preset number threshold;
[0119] A second number comparison unit is used to determine that the current risk level of the automated guided vehicle is a level 2 risk level if the number of human bodies in the current risk area is not less than a preset number threshold;
[0120] The time difference comparison unit is used to determine that the current risk level of the automatic guided vehicle is a third risk level if the time difference is less than a first preset time threshold.
[0121] In some specific implementations, the guided vehicle control module 13 includes:
[0122] A first speed reducing unit, configured to reduce the traveling speed of the automatic guided vehicle in the risk area according to a first preset ratio if the current risk level of the automatic guided vehicle is level one risk level;
[0123] A second speed reducing unit, configured to reduce the traveling speed of the automatic guided vehicle in the risk area according to a second preset ratio if the current risk level of the automatic guided vehicle is a level 2 risk level;
[0124] a route updating unit, configured to control the automatic guided vehicle to stop moving if the current risk level of the automatic guided vehicle is level three, and to update the moving route of the automatic guided vehicle in the risk area;
[0125] Correspondingly, the control device for adaptively adjusting the walking state of the AGV further includes:
[0126] The risk level revocation module is used to revoke the current risk level of the automatic guided vehicle and jump to the step of determining the corresponding target range according to the current position of the automatic guided vehicle and obtaining the point cloud data within the target range collected by the laser radar if it is determined that the automatic guided vehicle passes through the risk area.
[0127] Furthermore, the present application also discloses an electronic device. Figure 5 It is a structural diagram of an electronic device 20 shown according to an exemplary embodiment, and the content in the figure cannot be regarded as any limitation on the scope of use of this application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. Among them, the memory 22 is used to store a computer program, and the computer program is loaded and executed by the processor 21 to implement the relevant steps in the control method for adaptive adjustment of the walking state of the AGV disclosed in any of the aforementioned embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.
[0128] In this embodiment, the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs and is not specifically limited here.
[0129] In addition, the memory 22, as a carrier for storing resources, can be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon can include an operating system 221, a computer program 222, etc., and the storage method can be temporary storage or permanent storage.
[0130] The operating system 221 is used to manage and control the hardware devices and computer program 222 on the electronic device 20, which can be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program that can be used to complete the control method of adaptively adjusting the walking state of the AGV performed by the electronic device 20 disclosed in any of the aforementioned embodiments, the computer program 222 can further include a computer program that can be used to complete other specific tasks.
[0131] Furthermore, the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the control method for adaptively adjusting the walking state of the AGV disclosed above is implemented. The specific steps of the method can be referred to the corresponding contents disclosed in the above embodiments, and will not be repeated here.
[0132] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0133] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0134] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
[0135] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
[0136] The technical solution provided by the present application is introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for general technicians in this field, according to the idea of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.
Claims
1. A control method for adaptively adjusting the walking state of an AGV, characterized in that: include: Determine the corresponding target range according to the current position of the automatic guided vehicle, and obtain the point cloud data within the target range collected by the laser radar; Determine the data type of the point cloud data by using a preset deep learning model, and determine environmental information within the target range according to the data type of the point cloud data, so as to construct a risk area corresponding to the automated guided vehicle based on the environmental information; The current risk level of the automatic guided vehicle is determined based on the environmental information, and the walking state of the automatic guided vehicle is adjusted based on the risk level and the risk area.
2. The control method for adaptively adjusting the walking state of an AGV according to claim 1 is characterized in that: The determining the data type of the point cloud data by using a preset deep learning model, and determining the environmental information within the target range according to the data type of the point cloud data, includes: If the point cloud data is determined to be ground point cloud data through the preset deep learning model, a plane normal vector of the ground point cloud data is calculated to determine the inclination angle of the ground relative to the horizontal plane and the thickness of the ground point cloud according to the plane normal vector; Accordingly, determining the current risk level of the automatic guided vehicle based on the environmental information includes: If the tilt angle is greater than a preset tilt angle threshold, determining that the current risk level of the automated guided vehicle is a level one risk level; If the thickness of the ground point cloud is greater than a first preset thickness threshold and less than a second preset thickness threshold, determining that the current risk level of the automated guided vehicle is a secondary risk level; If the thickness of the ground point cloud is not less than the second preset thickness threshold, it is determined that the current risk level of the automatic guided vehicle is a third risk level.
3. The control method for adaptively adjusting the walking state of an AGV according to claim 1, characterized in that: The determining the data type of the point cloud data by using a preset deep learning model, and determining the environmental information within the target range according to the data type of the point cloud data, includes: If the point cloud data is determined to be ground point cloud data through the preset deep learning model, and it is determined that there is a preset sparse area in the ground point cloud data whose data density is less than a preset density, then it is determined that there is a first target area in the target range; If the point cloud data is determined to be ground point cloud data through the preset deep learning model, and it is determined that a preset hole area exists in the ground point cloud data, then it is determined that a second target area exists within the target range.
4. The control method for adaptively adjusting the walking state of an AGV according to claim 3 is characterized in that: The determining the current risk level of the automated guided vehicle based on the environmental information includes: When the first target area exists within the target range, determining a difference between a minimum value of a regional point cloud in the preset sparse area corresponding to the first target area and a value of a ground point cloud plane; If the difference is less than a preset difference threshold, determining that the risk level of the automated guided vehicle in the risk area is a secondary risk level; If the difference is not less than a preset difference threshold, determining that the risk level of the automated guided vehicle in the risk area is a third risk level; When the second target area exists within the target range, the risk level of the automatic guided vehicle in the risk area is directly determined to be a third risk level.
5. The control method for adaptively adjusting the walking state of an AGV according to claim 1, characterized in that: The determining the data type of the point cloud data by using a preset deep learning model, and determining the environmental information within the target range according to the data type of the point cloud data, includes: If the point cloud data is determined to be wall point cloud data through the preset deep learning model, then determine whether there is a lane to be passed within the target range, and if so, determine the distance value between the wall point cloud data, and determine the lane width of the lane to be passed according to the distance value; Accordingly, determining the current risk level of the automatic guided vehicle based on the environmental information includes: Determining a difference between the lane width and a maximum width value of the automated guided vehicle; If the difference is not less than the first preset distance threshold and less than the second preset distance threshold, determining that the risk level of the automated guided vehicle in the risk area is a first-level risk level; If the difference is not less than the third preset distance threshold and less than the first preset distance threshold, determining that the risk level of the automated guided vehicle in the risk area is a secondary risk level; If the distance is less than a third preset distance threshold, it is determined that the risk level of the automatic guided vehicle in the risk area is a third risk level.
6. The control method for adaptively adjusting the walking state of an AGV according to claim 1, characterized in that: The determining the data type of the point cloud data by using a preset deep learning model, and determining the environmental information within the target range according to the data type of the point cloud data, includes: If the point cloud data is determined to be human point cloud data by the preset deep learning model, the movement speed, the direction of travel and the number of human bodies within the target range are determined by the human point cloud data; Accordingly, determining the current risk level of the automatic guided vehicle based on the environmental information includes: Determining an encounter time between the human body and the automated guided vehicle based on the movement speed and the travel direction, and determining a time difference between the encounter time and a current time; If the number of human bodies in the current risk area is less than a preset number threshold, determining that the current risk level of the automated guided vehicle is a level one risk level; If the number of human bodies in the current risk area is not less than a preset number threshold, determining that the current risk level of the automated guided vehicle is a level 2 risk level; If the time difference is less than the first preset time threshold, it is determined that the current risk level of the automatic guided vehicle is a third risk level.
7. The control method for adaptively adjusting the walking state of an AGV according to any one of claims 1 to 6, characterized in that: The adjusting the walking state of the automatic guided vehicle based on the risk level and the risk area includes: If the current risk level of the automated guided vehicle is level one, reducing the walking speed of the automated guided vehicle in the risk area according to a first preset ratio; If the current risk level of the automated guided vehicle is level 2, reducing the walking speed of the automated guided vehicle in the risk area according to a second preset ratio; If the current risk level of the automated guided vehicle is level three, controlling the automated guided vehicle to stop moving, and updating the moving route of the automated guided vehicle in the risk area; Correspondingly, after adjusting the walking state of the automatic guided vehicle based on the risk level and the risk area, the method further includes: If it is determined that the automated guided vehicle passes through the risk area, the current risk level of the automated guided vehicle is canceled and the process jumps to the step of determining the corresponding target range according to the current position of the automated guided vehicle and acquiring the point cloud data within the target range collected by the laser radar.
8. A control device for adaptively adjusting the walking state of an AGV, characterized in that: include: A data collection module, used to determine the corresponding target range according to the current position of the automatic guided vehicle, and obtain point cloud data within the target range collected by the laser radar; A data processing module, used to determine the data type of the point cloud data through a preset deep learning model, and determine the environmental information within the target range according to the data type of the point cloud data, so as to construct a risk area corresponding to the automatic guided vehicle based on the environmental information; The guided vehicle control module is used to determine the current risk level of the automatic guided vehicle based on the environmental information, and adjust the walking state of the automatic guided vehicle based on the risk level and the risk area.