Intelligent shuttle vehicle obstacle avoidance method and system based on visual identification

Abstract

The invention provides an intelligent shuttle vehicle obstacle avoidance method and system based on visual identification, relates to the technical field of shuttle vehicle obstacle avoidance methods, and can quantitatively evaluate the obstacle shielding level, accurately judge the traffic capacity and the path congestion degree and realize dynamic adjustment and path optimization by introducing the safety coefficient and the risk index of a motion path. Compared with the prior art, the scheme has the advantages that the flexibility and adaptability of the system are improved, and complex and changing environments can be better coped with. Through the risk index updated in real time, the system can provide potential risk early warning to remind a user to clean a road in time, and long-term path management and efficient operation are ensured. On the whole, through the method, the passing efficiency and safety of the intelligent shuttle vehicle are greatly improved, and key problems in the prior art are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of shuttle vehicle obstacle avoidance methods, and specifically to an intelligent shuttle vehicle obstacle avoidance method and system based on visual recognition. Background Art

[0002] In modern industrial automation and warehouse management, smart shuttles are widely used in material handling and logistics distribution. However, in actual operation, these shuttles often encounter obstacles such as stacked goods and raw materials in the warehouse, which makes it impossible to pass through the preset path smoothly, increasing the risk of collision and low efficiency. Existing technologies mostly use simple distance judgment or occlusion ratio to avoid obstacles. This method lacks accuracy and flexibility in complex environments, easily leads to misjudgment and inability to adjust the path in time. It is still in the stage of real-time detection and is difficult to provide risk warnings.

[0003] For example, a dual-objective task multi-four-way shuttle path optimization and obstacle avoidance method and system disclosed in Chinese patent publication number CN118746982A includes obtaining basic information of a storage system and establishing a grid map of a four-way shuttle storage system; establishing a task model with the goal of minimizing the total time required for all four-way shuttles to complete a single-objective task; establishing a CCBS-EDT algorithm solution model, solving the task model, and performing path planning; wherein, the CCBS-EDT algorithm solution model uses the CCBS algorithm for initial global path planning, and when a conflict is detected, the SIPP algorithm is used to expand the node selection strategy, and time and space constraints are added, a two-label strategy and a minimum turning strategy are introduced, and the paths of conflicting vehicles are locally replanned; the replanned paths are re-detected for conflicts until a set of optimal paths without conflicts are found; different shuttles avoid obstacles according to the optimal paths.

[0004] Although it can plan the path of the shuttle's movement trajectory, it is still a passive detection, that is, the shuttle's path is re-planned according to the conflict position. It cannot provide risk warnings for the shuttle's subsequent movement, nor can it remind users, reducing its applicability in complex environments.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide an intelligent shuttle vehicle obstacle avoidance method and system based on visual recognition to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] An intelligent shuttle vehicle obstacle avoidance method based on visual recognition, the specific steps include:

[0009] S1: Collect the real-time position information and front image of the shuttle, generate obstacle avoidance information of the shuttle, and generate the occlusion level of the obstacle based on the obstacle avoidance information;

[0010] S2: judging whether the shuttle can pass based on the occlusion level of the obstacle and the risk index of the preset motion path, if it can pass, executing step S3, if it cannot pass, stopping the movement and issuing an alarm;

[0011] S3: Generate an obstacle avoidance path according to the obstacle avoidance information to temporarily replace the preset motion path for obstacle avoidance, and restore the preset motion path to continue moving after passing the obstacle;

[0012] S4: Repeat steps S1 to S3 until the end of the motion path is reached, and update the risk index of the motion path according to the occlusion levels of all obstacles on the motion path.

[0013] Preferably, the motion path is arranged to coincide with the center line of the road surface, and the obstacle avoidance information includes the distance between the shuttle vehicle and the obstacle, the occlusion width of the obstacle, and the remaining width of the motion path.

[0014] Preferably, the shuttle is equipped with an array light source, and the distance between the shuttle and the obstacle is measured by using an array light source projection method;

[0015] Then, the occlusion width of the obstacle is generated according to the distance between the shuttle and the obstacle. The generation logic is:

[0016] Input the front image into the pre-trained image recognition model for edge recognition and generate the pixel width of the obstacle in the front image;

[0017] The visual angle width of the obstacle in the front image is calculated based on the size information of the obstacle. The calculation method is:

[0018]

[0019] Where θ represents the viewing angle width, W pi represents pixel width, and f represents the focal length of the lens;

[0020] Then calculate the occlusion width of the obstacle according to the viewing angle width. The calculation method is:

[0021]

[0022] Where W ac represents the occlusion width, and D represents the distance between the shuttle and the obstacle;

[0023] The remaining width of the motion path is calculated as:

[0024] W sy =W z -W ac

[0025] Where W sy Indicates the remaining width, W z Indicates the preset road width.

[0026] Preferably, the logic for generating the occlusion level of an obstacle according to the obstacle avoidance information is:

[0027] The safety factor of the motion path is generated based on the obstacle avoidance information. The calculation method is:

[0028]

[0029] Where W cs represents the width of the shuttle, ξ represents the risk index of the motion path, ξ=1 in the initial state, C represents the discriminant factor, and its value is determined as follows:

[0030]

[0031] Then, the occlusion level of the obstacle is determined based on the safety factor of the motion path:

[0032] When , the obstacle is considered to be completely blocked and the shuttle cannot pass;

[0033] When , the obstacle is considered to be a serious obstruction and the shuttle can pass;

[0034] When , the obstacle is considered to be partially blocked and the shuttle can pass;

[0035] In the formula Indicates the preset safety threshold.

[0036] Preferably, the logic for generating an obstacle avoidance path according to the obstacle avoidance information is:

[0037] The lateral moving distance of the shuttle is calculated based on the obstacle avoidance information. The calculation method is:

[0038]

[0039] After moving horizontally, the shuttle continues to move forward at a constant speed, while collecting the side image of the shuttle close to the obstacle;

[0040] The side image is input into the pre-trained image recognition model for edge recognition, and the time when the edge image is recognized is marked as the reference time;

[0041] The remaining time for the shuttle to continue to move forward at a uniform speed is calculated based on the edge image and the reference time, and after the remaining time, the shuttle moves laterally in the opposite direction by the lateral movement distance to restore to the preset movement path.

[0042] Preferably, the remaining time generation logic is:

[0043] Starting from the reference time, N frames of side images are continuously collected, and the lateral coordinates of the edge boundary in the edge image corresponding to each side image are calculated. The calculation method is:

[0044]

[0045] In the formula Indicates the horizontal coordinate of the edge boundary, y i represents the horizontal coordinate of the i-th edge point on the edge boundary, and n represents the number of edge points on the edge boundary;

[0046] Detect the change in the horizontal coordinates of the edge boundary between two adjacent side images. The calculation method is: in represents the change in the horizontal coordinate of the edge boundary in the j-th side image, represents the lateral coordinate of the edge boundary in the j-th side image;

[0047] Then calculate the remaining time according to the change of the horizontal coordinates. The calculation method is:

[0048]

[0049] Where W tx Represents the image width of the edge image, t represents the interval time for collecting the side image, ∈ represents the preset scale factor, ∈∈[1.1,1.3].

[0050] Preferably, the risk index of the motion path is calculated according to the occlusion level of all obstacles on the motion path, and the calculation method is:

[0051]

[0052] Where τ(p) represents the occlusion index of the pth obstacle, and its value is obtained as follows:

[0053]

[0054] The preset risk index is then updated based on the calculated risk index.

[0055] An intelligent shuttle vehicle obstacle avoidance system based on visual recognition, the obstacle avoidance system is used to execute the above-mentioned obstacle avoidance method, comprising:

[0056] A GPS module, which is installed on the shuttle body and is used to collect real-time location information of the shuttle;

[0057] An array light source module, which is installed at the front of the shuttle and centered at the front of the shuttle, and is used to measure the distance between the shuttle and the obstacle;

[0058] A first camera module, which is installed at the front of the shuttle and centered at the front of the shuttle, and is used to capture a front image of the shuttle;

[0059] A second camera module, wherein the second camera module has two sets of cameras, which are respectively installed on both sides of the shuttle vehicle and are installed in the center of both sides of the shuttle vehicle, and are used to collect side images of the shuttle vehicle;

[0060] An image analysis module, wherein a trained image recognition model is preset in the image analysis module for edge recognition of the front image and the side image;

[0061] A data processing module, the data processing module is used to generate obstacle avoidance information of the shuttle vehicle and the occlusion level of the obstacle according to the position information of the shuttle vehicle and the front image;

[0062] A data judgment module, which is used to judge whether the shuttle can pass according to the occlusion level of the obstacle and the risk index of the preset motion path;

[0063] The comprehensive processing module is used to generate an obstacle avoidance path according to the obstacle avoidance information to temporarily replace the preset motion path for obstacle avoidance, and update the risk index of the motion path.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] By introducing the safety factor and risk index of the motion path, the present invention can quantitatively evaluate the obstacle occlusion level, accurately judge the traffic capacity and the degree of path congestion, and dynamically adjust and optimize the path. Compared with traditional technologies, this solution improves the flexibility and adaptability of the system and better copes with complex and changing environments. Through real-time updated risk indexes, the system can provide potential risk warnings, remind users to clear the road in time, and ensure long-term path management and efficient operation. Overall, this method greatly improves the traffic efficiency and safety of smart shuttles, and effectively solves key problems in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0067] Figure 2It is a schematic diagram of the module structure of the present invention. DETAILED DESCRIPTION

[0068] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0069] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0070] Example:

[0071] See also Figure 1 , the present invention provides a technical solution:

[0072] An intelligent shuttle vehicle obstacle avoidance method based on visual recognition, the specific steps include:

[0073] S1: Collect the real-time position information and front image of the shuttle vehicle, generate obstacle avoidance information about the shuttle vehicle, the obstacle avoidance information includes the distance between the shuttle vehicle and the obstacle, the occlusion width of the obstacle, the remaining width of the motion path, and generate the occlusion level of the obstacle based on the obstacle avoidance information.

[0074] The shuttle is equipped with an array light source. The distance between the shuttle and the obstacle is measured using the array light source projection method. Using an array light source to calculate the distance between the camera and the object is a common three-dimensional measurement technology, especially widely used in computer vision and industrial automation. This method usually projects a grid or stripe pattern onto the surface of the object, and the camera captures the deformation of these patterns. Through geometric transformation and triangulation, the depth information of the object surface can be calculated.

[0075] Then, the occlusion width of the obstacle is generated according to the distance between the shuttle and the obstacle. The generation logic is:

[0076] Input the front image into the pre-trained image recognition model for edge recognition and generate the pixel width of the obstacle in the front image;

[0077] The visual angle width of the obstacle in the front image is calculated based on the size information of the obstacle. The calculation method is:

[0078]

[0079] Where θ represents the viewing angle width, W pi represents pixel width, and f represents the focal length of the lens;

[0080] Then calculate the occlusion width of the obstacle according to the viewing angle width. The calculation method is:

[0081]

[0082] Where W ac represents the occlusion width, and D represents the distance between the shuttle and the obstacle;

[0083] The remaining width of the motion path is calculated as:

[0084] W sy =W z -W ac

[0085] Where W sy Indicates the remaining width, W z Indicates the preset road width.

[0086] In this step, since the obstacle is still far away from the shuttle when the front image detects the existence of an obstacle, the distance between the obstacle and the shuttle is taken into consideration, and the viewing angle width method is used to calculate the occlusion width of the obstacle. When the object is far away, this method can better reflect the actual situation.

[0087] It is understandable that in factory warehouses and other places, when goods are stacked, the road may be occupied due to factors such as not being placed in place or too many raw materials, which may cause collisions when the shuttle moves according to the preset movement path. The blocked width reflects the width of the road that is occupied, and similarly, the remaining width reflects the width of the remaining passable part of the road.

[0088] The logic of generating the occlusion level of obstacles based on obstacle avoidance information is:

[0089] The safety factor of the motion path is generated based on the obstacle avoidance information. The calculation method is:

[0090]

[0091] Where W cs represents the width of the shuttle, ξ represents the risk index of the motion path, ξ=1 in the initial state, C represents the discriminant factor, and its value is determined as follows:

[0092]

[0093] The occlusion level of the obstacle is then determined based on the safety factor of the motion path.

[0094] S2: Based on the occlusion level of the obstacle and the risk index of the preset motion path, determine whether the shuttle can pass. The motion path is set to coincide with the center line of the road surface. If it can pass, execute step S3. If it cannot pass, stop moving and issue an alarm.

[0095] When , the obstacle is considered to be completely blocked and the shuttle cannot pass;

[0096] When , the obstacle is considered to be a serious obstruction and the shuttle can pass;

[0097] When , the obstacle is considered to be partially blocked and the shuttle can pass;

[0098] In the formula Indicates the preset safety threshold.

[0099] It can be seen from the calculation formula of the safety factor of the motion path that if the remaining width of the road surface is less than the body width of the shuttle vehicle, then the overall safety factor is negative, indicating that the road surface cannot be passed; conversely, if the remaining width of the road surface is greater than or equal to the body width of the shuttle vehicle, then the overall safety factor is positive, indicating that although the road surface is blocked, it can still be passed, and the larger the remaining width of the road surface, the lower the obstruction level of the obstacle, and the easier it is for the shuttle vehicle to pass.

[0100] In this step, by introducing the safety factor of the motion path and its calculation method, the obstacle occlusion level can be quantitatively evaluated, which is more accurate than a simple binary judgment (can pass / cannot pass), can better reflect the actual situation, and can better dynamically adjust the shuttle's obstacle avoidance behavior, which helps to cope with complex and changing environments and improve the flexibility and adaptability of the system.

[0101] S3: Generate an obstacle avoidance path according to the obstacle avoidance information to temporarily replace the preset motion path for obstacle avoidance, and restore to the preset motion path to continue moving after passing the obstacle.

[0102] The logic of generating an obstacle avoidance path based on obstacle avoidance information is:

[0103] The lateral moving distance of the shuttle is calculated based on the obstacle avoidance information. The calculation method is:

[0104]

[0105] After moving horizontally, the shuttle continues to move forward at a constant speed, while collecting the side image of the shuttle close to the obstacle;

[0106] The side image is input into the pre-trained image recognition model for edge recognition, and the time when the edge image is recognized is marked as the reference time;

[0107] The remaining time for the shuttle to continue to move forward at a constant speed is calculated based on the edge image and the reference time, and the shuttle moves laterally in the opposite direction at a lateral moving distance after the remaining time to restore to the preset motion path.

[0108] It can be understood that in the initial movement path, the shuttle moves along the center line of the road surface, while in the obstacle avoidance path, the shuttle moves along the center line of the part of the road surface that is not blocked by obstacles, and then moves after slowing down. This is to avoid the shuttle from colliding with goods, raw materials, etc. stacked on both sides of the road surface during the obstacle avoidance process, thereby improving the safety of the obstacle avoidance path.

[0109] The remaining time generation logic is:

[0110] Starting from the reference time, N frames of side images are continuously collected, and the lateral coordinates of the edge boundary in the edge image corresponding to each side image are calculated. The calculation method is:

[0111]

[0112] In the formula Indicates the horizontal coordinate of the edge boundary, y i represents the horizontal coordinate of the i-th edge point on the edge boundary, and n represents the number of edge points on the edge boundary;

[0113] Detect the change in the horizontal coordinates of the edge boundary between two adjacent side images. The calculation method is: in represents the change in the horizontal coordinate of the edge boundary in the j-th side image, represents the lateral coordinate of the edge boundary in the j-th side image;

[0114] Then calculate the remaining time according to the change of the horizontal coordinates. The calculation method is:

[0115]

[0116] Where W tx Represents the image width of the edge image, t represents the interval time for collecting the side image, ∈ represents the preset scale factor, ∈∈[1.1,1.3].

[0117] The shuttle's obstacle avoidance path can be divided into three major stages. The first step is to move laterally to avoid obstacles; the second step is to continue moving along the road surface and cross the obstacles; the third step is to move laterally in the opposite direction to return to the initial movement path, which is the center line of the road surface.

[0118] In the second step, the motion trajectory of the shuttle can be further divided into three time periods: the first period is when the shuttle does not cross the obstacle, and the side of the shuttle is blocked by the obstacle, so there is no edge boundary in the side image; the second period is when the shuttle body part crosses the obstacle, and part of the side of the shuttle crosses the obstacle, and the side image detects the edge of the obstacle, so the edge boundary is also detected, and this time is marked as the reference time; the third period is the remaining time calculated based on the side image after continuously collecting N frames of side images. Therefore, in the second step, the total motion time of the shuttle should be the time from the end of the lateral movement to the reference time, plus the time Nt for collecting the side image, plus the remaining time. The proportional factor is set in the remaining time because the camera on the side of the shuttle is also set in the center. When the obstacle leaves the camera's shooting range, the tail of the shuttle may still overlap with the obstacle. Therefore, a proportional factor is added to increase the remaining time, so that the shuttle can move in the opposite direction after the remaining time to return to the initial motion path to avoid the tail of the shuttle from colliding with the obstacle. The proportional factor should not be set too large to avoid the shuttle vehicle not having enough space to turn and make the next lateral movement when it encounters a new obstacle. Therefore, the specific value needs to be adjusted by the user according to the actual situation.

[0119] S4: Repeat steps S1 to S3 until the end of the motion path is reached, and update the risk index of the motion path according to the occlusion levels of all obstacles on the motion path.

[0120] The risk index of the motion path is calculated according to the occlusion level of all obstacles on the motion path. The calculation method is:

[0121]

[0122] Where τ(p) represents the occlusion index of the pth obstacle, and its value is obtained as follows:

[0123]

[0124] The preset risk index is then updated based on the calculated risk index.

[0125] It can be seen from the calculation formula of the safety factor of the motion path that the safety factor reflects the degree of occlusion of a single obstacle, while the risk index reflects the degree of occlusion of all obstructions on this motion path. In the initial case, the risk factor is set to 1, which will not affect the value of the safety factor. After the shuttle moves along the motion path once, the risk index of the motion path will be updated according to the occlusion level of each obstruction. The more obstacles there are on the motion path, the higher the occlusion level of the obstacles, and the higher the risk index of the motion path. Correspondingly, when the shuttle arrives at the same obstacle for the second time, its safety factor will decrease, and each time it completes a motion path and arrives at the same obstacle again, its safety factor will continue to decrease until the obstacles in the motion path are cleared. During this period, the safety factor of the shuttle is still greater than 0, which means that it does not affect normal traffic, but it will increase the occlusion level of obstacles to remind users to clear the road in time. In other words, if the shuttle vehicle has fewer transportation times but the risk index of the movement path is higher, it means that although the movement path can be passed, the road is very congested; if the shuttle vehicle has more transportation times but the risk index of the movement path is higher, it means that although the road is only generally congested, there are too many transportation times and a collision may occur at any time, and the potential risk is higher. In both cases, users need to clear the road in a timely manner.

[0126] In this step, by updating the risk index, the cumulative impact of obstacles on the path is reflected in real time, allowing the system to dynamically adjust and adapt to environmental changes, improve the efficiency and safety of the path, and at the same time provide potential risk warnings to remind users to clear the road in time, which is helpful for long-term path management and efficiency improvement.

[0127] In summary, the present invention can quantitatively evaluate the obstacle occlusion level, accurately judge the traffic capacity and the degree of path congestion, and dynamically adjust and optimize the path by introducing the safety factor and risk index of the motion path. Compared with traditional technologies, this solution improves the flexibility and adaptability of the system and better copes with complex and changing environments. Through real-time updated risk indexes, the system can provide potential risk warnings, remind users to clear roads in time, and ensure long-term path management and efficient operation. Overall, this method greatly improves the traffic efficiency and safety of smart shuttles, and effectively solves key problems in the prior art.

[0128] See also Figure 2 The present invention also provides an intelligent shuttle vehicle obstacle avoidance system based on visual recognition, which is used to execute the above obstacle avoidance method, including:

[0129] GPS module, which is installed on the shuttle body and is used to collect the real-time location information of the shuttle;

[0130] An array light source module is installed at the front of the shuttle and is installed in the center of the front of the shuttle to measure the distance between the shuttle and the obstacle.

[0131] A first camera module, the first camera module is installed at the front of the shuttle vehicle and is installed in the center of the front of the shuttle vehicle, and is used to capture the front image of the shuttle vehicle;

[0132] The second camera module has two sets of cameras, which are respectively installed on both sides of the shuttle vehicle and are installed in the center of both sides of the shuttle vehicle to collect side images of the shuttle vehicle;

[0133] An image analysis module, in which a trained image recognition model is preset for edge recognition of front and side images;

[0134] A data processing module, the data processing module is used to generate obstacle avoidance information of the shuttle vehicle and the occlusion level of the obstacle according to the position information of the shuttle vehicle and the front image;

[0135] A data judgment module is used to judge whether the shuttle can pass according to the occlusion level of the obstacle and the risk index of the preset motion path;

[0136] The comprehensive processing module is used to generate an obstacle avoidance path according to the obstacle avoidance information to temporarily replace the preset motion path for obstacle avoidance, and update the risk index of the motion path.

[0137] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0138] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0139] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0140] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. An intelligent shuttle vehicle obstacle avoidance method based on visual recognition, characterized in that: The specific steps include: S1: Collect the real-time position information and front image of the shuttle, generate obstacle avoidance information of the shuttle, and generate the occlusion level of the obstacle based on the obstacle avoidance information; S2: judging whether the shuttle can pass based on the occlusion level of the obstacle and the risk index of the preset motion path, if it can pass, executing step S3, if it cannot pass, stopping the movement and issuing an alarm; S3: Generate an obstacle avoidance path according to the obstacle avoidance information to temporarily replace the preset motion path for obstacle avoidance, and restore the preset motion path to continue moving after passing the obstacle; S4: Repeat steps S1 to S3 until the end of the motion path is reached, and update the risk index of the motion path according to the occlusion levels of all obstacles on the motion path.

2. According to claim 1, a method for avoiding obstacles of an intelligent shuttle vehicle based on visual recognition is characterized in that: The motion path is arranged to coincide with the center line of the road surface, and the obstacle avoidance information includes the distance between the shuttle vehicle and the obstacle, the occlusion width of the obstacle, and the remaining width of the motion path.

3. The intelligent shuttle vehicle obstacle avoidance method based on visual recognition according to claim 2 is characterized in that: The shuttle is equipped with an array light source, and the distance between the shuttle and the obstacle is measured by using an array light source projection method; Then, the occlusion width of the obstacle is generated according to the distance between the shuttle and the obstacle. The generation logic is: Input the front image into the pre-trained image recognition model for edge recognition and generate the pixel width of the obstacle in the front image; The visual angle width of the obstacle in the front image is calculated based on the size information of the obstacle. The calculation method is: Where θ represents the viewing angle width, W pi represents pixel width, and f represents the focal length of the lens; Then calculate the occlusion width of the obstacle according to the viewing angle width. The calculation method is: Where W ac represents the occlusion width, and D represents the distance between the shuttle and the obstacle; The remaining width of the motion path is calculated as: IN sy =In z -IN ac Where W sy Indicates the remaining width, W z Indicates the preset road width.

4. The intelligent shuttle vehicle obstacle avoidance method based on visual recognition according to claim 3 is characterized in that: The logic of generating the occlusion level of obstacles based on obstacle avoidance information is: The safety factor of the motion path is generated based on the obstacle avoidance information. The calculation method is: Where W cs represents the width of the shuttle, ξ represents the risk index of the motion path, ξ=1 in the initial state, C represents the discriminant factor, and its value is determined as follows: Then, the occlusion level of the obstacle is determined based on the safety factor of the motion path: When , the obstacle is considered to be completely blocked and the shuttle cannot pass; When , the obstacle is considered to be a serious obstruction and the shuttle can pass; When , the obstacle is considered to be partially blocked and the shuttle can pass; In the formula Indicates the preset safety threshold.

5. The intelligent shuttle vehicle obstacle avoidance method based on visual recognition according to claim 4 is characterized in that: The logic of generating an obstacle avoidance path based on obstacle avoidance information is: The lateral moving distance of the shuttle is calculated based on the obstacle avoidance information. The calculation method is: After moving horizontally, the shuttle continues to move forward at a constant speed, while collecting the side image of the shuttle close to the obstacle; The side image is input into the pre-trained image recognition model for edge recognition, and the time when the edge image is recognized is marked as the reference time; The remaining time for the shuttle to continue to move forward at a uniform speed is calculated based on the edge image and the reference time, and after the remaining time, the shuttle moves laterally in the opposite direction by the lateral movement distance to restore to the preset movement path.

6. The method for avoiding obstacles of an intelligent shuttle vehicle based on visual recognition according to claim 5, characterized in that: The generation logic of the remaining time is: Starting from the reference time, N frames of side images are continuously collected, and the lateral coordinates of the edge boundary in the edge image corresponding to each side image are calculated. The calculation method is: In the formula Indicates the horizontal coordinate of the edge boundary, y i represents the horizontal coordinate of the i-th edge point on the edge boundary, and n represents the number of edge points on the edge boundary; Detect the change in the horizontal coordinates of the edge boundary between two adjacent side images. The calculation method is: in represents the change in the horizontal coordinate of the edge boundary in the j-th side image, represents the lateral coordinate of the edge boundary in the j-th side image; Then calculate the remaining time according to the change of the horizontal coordinates. The calculation method is: Where W tx Represents the image width of the edge image, t represents the interval time for collecting the side image, ∈ represents the preset scale factor, ∈∈[1.1,1.3].

7. The intelligent shuttle vehicle obstacle avoidance method based on visual recognition according to claim 6 is characterized in that: The risk index of the motion path is calculated according to the occlusion level of all obstacles on the motion path. The calculation method is: Where τ(p) represents the occlusion index of the pth obstacle, and its value is obtained as follows: The preset risk index is then updated based on the calculated risk index.

8. An intelligent shuttle vehicle obstacle avoidance system based on visual recognition, characterized in that: The obstacle avoidance system is used to execute the obstacle avoidance method according to any one of claims 1 to 7, comprising: A GPS module, which is installed on the shuttle body and is used to collect real-time location information of the shuttle; An array light source module, which is installed at the front of the shuttle and centered at the front of the shuttle, and is used to measure the distance between the shuttle and the obstacle; A first camera module, which is installed at the front of the shuttle and centered at the front of the shuttle, and is used to capture a front image of the shuttle; A second camera module, wherein the second camera module has two sets of cameras, which are respectively installed on both sides of the shuttle vehicle and are installed in the center of both sides of the shuttle vehicle, and are used to collect side images of the shuttle vehicle; An image analysis module, wherein a trained image recognition model is preset in the image analysis module for edge recognition of the front image and the side image; A data processing module, the data processing module is used to generate obstacle avoidance information of the shuttle vehicle and the occlusion level of the obstacle according to the position information of the shuttle vehicle and the front image; A data judgment module, which is used to judge whether the shuttle can pass according to the occlusion level of the obstacle and the risk index of the preset motion path; The comprehensive processing module is used to generate an obstacle avoidance path according to the obstacle avoidance information to temporarily replace the preset motion path for obstacle avoidance, and update the risk index of the motion path.

Images