An Obstacle Avoidance Method for Logistics Carts Based on Machine Vision and Magnetic Field Potential Energy

Abstract

The present invention discloses an obstacle avoidance method for a logistics trolley based on machine vision and magnetic field potential energy, which includes the following steps: S1. Perform path planning through the artificial potential field method, and the central computer sends a working instruction to make the AGV trolley reach the designated cargo position; S2. The central computer determines whether the AGV trolley has completed the operation within the preset time; S3. If the AGV trolley has not completed the operation within the preset time, transmit the information of the identified obstacle and the blocked AGV trolley to the central computer through the WiFi communication module of the AGV trolley; S4. The central computer issues an order to make the AGV trolley move passively; S5. Determine again whether the AGV trolley has completed the operation within the preset time; S6. If it does not reach the designated cargo position within the specified time, abandon the use of the artificial potential field method to plan the path, generate a magnetic field through the electromagnet of the trolley to control the trolley, so as to re-plan the route and prevent the AGV trolley from falling into a local optimum solution.

Description

Technical Field

[0001] The present invention belongs to the technical field of logistics transportation, and particularly relates to an obstacle avoidance method for a logistics trolley based on machine vision and magnetic field potential energy. Background Art

[0002] With the problem of unmanned AGV trolley scheduling becoming a hot topic in the logistics industry, the dynamic obstacle avoidance method for AGV trolleys has made great progress. Existing methods for realizing the dynamic obstacle avoidance of AGV trolleys by the artificial potential field method are mostly limited to generating virtual gravitational and repulsive forces by improving the algorithm or combining other algorithms. However, the improved algorithms still have more or less problems, such as: (1) when the object is far from the target point, the gravitational force will become extremely large, and the relatively small repulsive force can be even negligible, making it impossible to avoid obstacles; (2) when there are obstacles near the target point, the repulsive force will be very large, and the gravitational force is relatively small, making it difficult for the object to reach the target point; (3) at a certain point, the gravitational force and the repulsive force are just equal in magnitude and opposite in direction, then the object is likely to fall into a local optimal solution or oscillation.

[0003] Disadvantages of the existing traditional artificial potential field method:

[0004] (1) When the object is far from the target point, the gravitational force will become extremely large, and the relatively small repulsive force can be even negligible, making it impossible to avoid obstacles.

[0005] (2) When there are obstacles near the target point, the repulsive force will be very large, and the gravitational force is relatively small, making it difficult for the object to reach the target point;

[0006] (3) At a certain point, the gravitational force and the repulsive force are just equal in magnitude and opposite in direction, then the object is likely to fall into a local optimal solution or oscillation.

[0007] In the existing logistics field, magnetism is mainly applied to the goods carried by fixed AGV trolleys to prevent the goods from being displaced and falling, and there are few patented technologies that apply magnetism to dynamic and static obstacle avoidance. Summary of the Invention

[0008] The present invention overcomes the deficiencies of the prior art. In view of the above deficiencies, the present invention provides an obstacle avoidance method for a logistics trolley based on machine vision and magnetic field potential energy, and solves the above problems through targeted improvement and optimization.

[0009] The present invention provides the following technical solutions:

[0010] An obstacle avoidance method for a logistics trolley based on machine vision and magnetic field potential energy, comprising the following steps:

[0011] S1. Perform path planning by the artificial potential field method, and the central computer sends a work instruction to make the AGV trolley reach the specified cargo location;

[0012] S2. The central computer combines with the infrared sensor at the goods location to determine whether the AGV vehicle completes the operation within the preset time;

[0013] S3. If the AGV vehicle does not complete the operation within the preset time, then in combination with the yolov3 object detection algorithm, after identifying the obstacles encountered by the AGV vehicle through the camera of the AGV vehicle, the information of the identified obstacles and the blocked AGV vehicle is transmitted to the central computer through the WiFi communication module of the AGV vehicle;

[0014] S4. The central computer issues a command to change the repulsive force between the AGV vehicle and the obstacle by increasing the resistance of the rheostat of the AGV vehicle, so that the AGV vehicle moves passively;

[0015] S5. Determine again whether the AGV vehicle completes the operation within the preset time;

[0016] S6. If it does not reach the designated goods location within the specified time, then abandon using the artificial potential field method to plan the path, and control the vehicle through the electromagnet of the AGV vehicle to generate a magnetic field to re-plan the route.

[0017] More preferably, the central computer uses common obstacles as the training set of static and dynamic obstacles and puts them into the yolov3 algorithm for training, where the common obstacles include AGV vehicles, dropped goods, and static obstacles in the cargo hold.

[0018] Preferably, in S3, the information of the blocked AGV vehicle includes the label and position of the blocked AGV vehicle.

[0019] Preferably, in S4, after the AGV vehicle moves passively, the artificial potential field method jumps out of the local optimum or oscillation.

[0020] Preferably, in S5, if the AGV vehicle reaches the designated goods location within the specified time, the AGV vehicle performs the loading and unloading work.

[0021] Preferably, in S5, if the AGV vehicle reaches the designated goods location within the specified time, the AGV vehicle performs the loading and unloading work.

[0022] Preferably, in S6, the central computer calculates the magnetic force required to be generated by the electromagnet on each AGV vehicle, controls the current magnitude by using a rheostat, and changes the magnetic force generated by each electromagnet to push the AGV vehicle to the target position.

[0023] In view of the prior art, the logistics vehicle obstacle avoidance method based on machine vision and magnetic field potential energy provided by the present invention has the following

[0024] Beneficial effects:

[0025] 1. The present invention predicts the expected arrival time of an AGV cart by combining the path of the AGV cart with its traveling speed and reserving time, and determines whether to use machine vision to jump out of the possible local optimal solution.

[0026] 2. The present invention generates different magnetic forces by controlling the current magnitude to produce virtual gravitational force and virtual repulsive force applied in the traditional artificial potential field method for path planning and obstacle avoidance. The algorithm is materialized into physical objects, making it more operable. When an object is far from the target point, the central computer controls the rheostat to increase the repulsive force for obstacle avoidance; when there is an obstacle near the target point, the central computer controls the rheostat to decrease the repulsive force so that the AGV cart can reach the target point.

[0027] 3. The present invention combines the improved artificial potential field method, machine vision, and materialized artificial potential field method for static and dynamic obstacle avoidance. The three-layer guarantee ensures that the AGV cart can reach the designated cargo position and corrects the defects of the traditional artificial potential field method in generating local optimal solutions and oscillations, improving the operation ability of the AGV cart. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the following drawings.

[0029] Figure 1 is a flowchart of the obstacle avoidance method for a logistics cart based on machine vision and magnetic field potential energy of the present invention;

[0030] Figure 2 is a path diagram of a preferred embodiment of the present invention;

[0031] Figure 3 is a diagram showing getting stuck in a local optimum of a preferred embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] The following further describes in detail a method for an obstacle avoidance of a logistics cart based on machine vision and magnetic field potential energy in combination with specific embodiments. These embodiments are only for the purposes of comparison and explanation, and the present invention is not limited to these embodiments.

[0033] Embodiment

[0034] The method for an obstacle avoidance of a logistics cart based on machine vision and magnetic field potential energy provided in the embodiment of the present invention, as Figure 1 shown, includes the following steps:

[0035] S1. Perform path planning by the artificial potential field method, and the central computer sends a working instruction to make the AGV cart reach the designated cargo position;

[0036] S2. The central computer combines with the infrared sensor at the goods location to determine whether the AGV vehicle completes the operation within the preset time;

[0037] S3. If the AGV vehicle does not complete the operation within the preset time, then in combination with the yolov3 object detection algorithm, after identifying the obstacles encountered by the AGV vehicle through the camera of the AGV vehicle, the information of the identified obstacles and the blocked AGV vehicle is transmitted to the central computer through the WiFi communication module of the AGV vehicle;

[0038] S4. The central computer issues a command to change the repulsive force between the AGV vehicle and the obstacle by increasing the resistance of the rheostat of the AGV vehicle, causing the AGV vehicle to move passively;

[0039] S5. Determine again whether the AGV vehicle completes the operation within the preset time;

[0040] S6. If it does not reach the designated goods location within the specified time, then abandon using the artificial potential field method to plan the path, and control the vehicle through the electromagnet of the AGV vehicle to generate a magnetic field to re-plan the route.

[0041] More preferably, the central computer uses common obstacles as the training set of static and dynamic obstacles and puts them into the yolov3 algorithm for training, where the common obstacles include AGV vehicles, dropped goods, and static obstacles in the cargo hold.

[0042] Preferably, in S3, the information of the blocked AGV vehicle includes the label and position of the blocked AGV vehicle.

[0043] Preferably, in S4, after the AGV vehicle moves passively, the artificial potential field method jumps out of the local optimum or oscillation.

[0044] Preferably, in S5, if the AGV vehicle reaches the designated goods location within the specified time, the AGV vehicle performs the loading and unloading work.

[0045] Preferably, in S5, if the AGV vehicle reaches the designated goods location within the specified time, the AGV vehicle performs the loading and unloading work.

[0046] Preferably, in S6, the central computer calculates the magnetic force required to be generated by the electromagnet on each AGV vehicle, controls the current magnitude using the rheostat, and changes the magnetic force generated by each electromagnet to push the AGV vehicle to the target position.

[0047] In an embodiment, the AGV vehicle is equipped with a WiFi communication module, a camera, a rheostat, and an electromagnet, as Figure 2As shown in the figure, the AGV cart receives the work instruction from the central computer and travels from storage location No. 9 to storage location No. 112. When the infrared sensor at storage location No. 112 determines that the AGV cart has not arrived, that is, it has not arrived within the preset time. The preset value = (estimated distance / cart speed) + reserved time. The central computer judges whether the AGV cart has reached the designated storage location by predicting the arrival time and combining the infrared sensor at the storage location; if Figure 3 As shown in the figure, in the case of falling into a local optimum, the AGV cart fails to reach the storage location in time. After identifying the obstacles encountered by the AGV cart by combining the yolov3 object detection algorithm, the information of the obstacles identified by the AGV cart and the blocked AGV cart is transmitted to the central computer through the WiFi communication module. The central computer issues an order to change the repulsive force between the AGV cart and the obstacles by increasing the resistance of the rheostat of the AGV cart, so that the AGV cart is displaced passively and then the improved artificial potential field method is used to plan the route again.

[0048] When the improved artificial potential field method is still used to plan the route and falls into a local optimum solution and cannot reach the storage location in time, the artificial potential field method is abandoned for obstacle avoidance. At this time, the attractive force of the dynamic target point position on the AGV cart and the repulsive force of the obstacle on the AGV cart are used for obstacle avoidance, that is, the electromagnet of the AGV cart is used to generate a magnetic field to control the cart to re-plan the route.

[0049] The central computer calculates the magnetic force required to be generated by the electromagnets on each AGV cart, and uses the rheostat to control the current size to change the magnetic force generated by each electromagnet to push the AGV cart to the target position.

[0050] The obstacle avoidance method for the logistics cart based on machine vision and magnetic field potential energy provided by the above embodiments of the present invention calculates the path length through any improved artificial potential field method to plan the route, and combines the driving speed of the AGV cart with the reserved time to verify whether the AGV cart arrives at the storage location for operation within the estimated time; if it times out, machine vision is combined to judge the obstacles encountered by the AGV cart, and the repulsive force between the two is increased to make the AGV cart displaced passively and then the artificial potential field method is used again for path planning to achieve global optimization and avoid the AGV cart falling into a local optimum solution. The size of the magnetic force generated by the electromagnets at each label position is controlled by controlling the current size with a resistor, and the virtual attractive force and repulsive force that are difficult to change dynamically in the artificial potential field method are transformed into dynamic physical attractive force and repulsive force generated by the electromagnets; path planning is realized through the attractive force between the AGV cart and the storage location, dynamic obstacle avoidance is realized through the repulsive force between AGV carts, and static obstacle avoidance is realized through the repulsive force between the AGV cart and static obstacles.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting the protection scope of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A method for obstacle avoidance of a logistics trolley based on machine vision and magnetic field potential energy, characterized in that, It includes the following steps: S1. Perform path planning by the artificial potential field method, and the central computer sends a work instruction to make the AGV reach the specified cargo location; S2. The central computer combines the infrared sensor at the cargo location to determine whether the AGV completes the operation within the preset time; S3. If the AGV does not complete the operation within the preset time, then in combination with the yolov3 object detection algorithm, after identifying the obstacles encountered by the AGV through the camera of the AGV, the information of the identified obstacles and the blocked AGV is transmitted to the central computer through the WiFi communication module of the AGV; S4. The central computer issues a command to change the repulsive force between the AGV and the obstacle by increasing the resistance of the rheostat of the AGV, causing the AGV to move passively; S5. Judge again whether the AGV completes the operation within the preset time; S6. If it does not reach the specified cargo location within the specified time, abandon the use of the artificial potential field method to plan the path, and control the vehicle through the electromagnet of the AGV to generate a magnetic field to re-plan the route.

2. The method for obstacle avoidance of a logistics trolley based on machine vision and magnetic field potential energy according to claim 1, characterized in that, In S2, if the AGV completes the operation within the preset time, arrange the AGV to return to the specified parking position or execute the next work instruction issued by the central computer.

3. The method for obstacle avoidance of a logistics trolley based on machine vision and magnetic field potential energy according to claim 1, characterized in that, In S3, the fact that the AGV does not complete the operation within the preset time indicates that a local optimal solution has occurred or oscillations have occurred, and the central computer re-plans the route.

4. The method for obstacle avoidance of a logistics trolley based on machine vision and magnetic field potential energy according to claim 3, characterized in that, The central computer uses common obstacles as the training set of static and dynamic obstacles and puts them into the yolov3 algorithm for training, where the common obstacles include AGVs, dropped goods, and static obstacles in the cargo hold.

5. The method for obstacle avoidance of a logistics trolley based on machine vision and magnetic field potential energy according to claim 1, characterized in that, In S3, the information of the blocked AGV includes the label and position of the blocked AGV.

6. The method for obstacle avoidance of a logistics trolley based on machine vision and magnetic field potential energy according to claim 1, characterized in that, In S4, after the AGV moves passively, the artificial potential field method jumps out of the local optimal solution or oscillations.

7. The method for obstacle avoidance of a logistics trolley based on machine vision and magnetic field potential energy according to claim 1, characterized in that, In S5, if the AGV reaches the specified cargo location within the specified time, the AGV performs loading and unloading work.

8. The method for obstacle avoidance of a logistics trolley based on machine vision and magnetic field potential energy according to claim 1, characterized in that, In S6, the central computer calculates the magnetic force required to be generated by the electromagnets on each AGV, controls the current magnitude using the rheostat, and changes the magnetic force generated by each electromagnet to push the AGV to the target position.

Images