Obstacle avoidance method and system for automatic driving logistics trolley in agricultural area

Abstract

The invention relates to the technical field of automatic obstacle avoidance, and discloses an obstacle avoidance method and system for an automatic driving logistics trolley in an agricultural district, and the method comprises the steps: obtaining a driving path of the logistics trolley, and determining an original path image of the logistics trolley; video data, radar data and driving data acquired by a monitoring assembly in the driving process of the logistics trolley are acquired; based on the original path image and the video data, determining whether there is an obstacle, and if there is an obstacle, determining the type of the obstacle according to the video data and the radar data; determining a collision risk value of the logistics trolley and the obstacle according to the radar data, the driving data and the obstacle type, and further determining a collision risk level; and setting an obstacle avoidance instruction according to the collision risk level, and performing control to realize obstacle avoidance. According to the invention, through multi-sensor data fusion and collision risk dynamic evaluation, accurate obstacle avoidance decision is realized, and the safety and reliability of the agricultural region logistics trolley are improved.

Description

TECHNICAL FIELD

[0001] The application relates to an automatic obstacle avoidance technology field, in particular to an obstacle avoidance method and system for an automatic driving logistics small vehicle in a rural area. BACKGROUND

[0002] With the development of agricultural modernization, the demand for rural logistics transportation is increasing, and automatic driving logistics small vehicles are gradually applied to rural logistics transportation due to their high efficiency and low cost. However, the rural environment is complex, and there are dynamic obstacles (such as pedestrians and animals) and static obstacles (such as stones and debris), which puts high requirements on the obstacle avoidance capability of the logistics small vehicle. The existing obstacle avoidance methods mostly rely on single sensor data, lack accurate judgment of obstacle types and multi-dimensional evaluation of collision risks, resulting in inaccurate obstacle avoidance decisions, easy misjudgment or slow reaction, and affecting the driving safety and efficiency of the logistics small vehicle. Therefore, an obstacle avoidance method for an automatic driving logistics small vehicle in a rural area is urgently needed to realize automatic obstacle avoidance. SUMMARY

[0003] The application aims to provide an obstacle avoidance method for an automatic driving logistics small vehicle in a rural area, which aims to solve the problem that the existing obstacle avoidance methods mostly rely on single sensor data, lack accurate judgment of obstacle types and multi-dimensional evaluation of collision risks, resulting in inaccurate obstacle avoidance decisions, easy misjudgment or slow reaction, and affecting the driving safety and efficiency of the logistics small vehicle.

[0004] The application provides an obstacle avoidance method for an automatic driving logistics small vehicle in a rural area, which comprises the following steps:

[0005] Obtaining a plurality of destinations of a logistics small vehicle, formulating a driving path of the logistics small vehicle according to the destinations, and determining an original path image of the logistics small vehicle according to the driving path;

[0006] Controlling the logistics small vehicle to drive according to the driving path, and obtaining video data, radar data and driving data obtained by a monitoring component of the logistics small vehicle;

[0007] Based on the original path image, analyzing the video data to determine whether there is an obstacle, and if there is an obstacle, determining an obstacle type according to the video data and the radar data, wherein the obstacle type includes dynamic obstacles and static obstacles;

[0008] Determining a collision risk value of the logistics small vehicle and the obstacle according to the radar data, the driving data and the obstacle type, and determining a collision risk level according to the collision risk value;

[0009] Setting an obstacle avoidance instruction according to the collision risk level, and controlling the logistics small vehicle based on the obstacle avoidance instruction to realize obstacle avoidance.

[0010] Preferably, the driving path of the logistics trolley is determined according to the destination, comprising:

[0011] Obtaining the current position information of the logistics trolley, and generating a plurality of driving paths traversing all destinations by using a path planning algorithm;

[0012] Analyzing the plurality of driving paths to determine the driving distance of each driving path;

[0013] Comparing the driving distances to determine the minimum value of the driving distance, and setting the driving path corresponding to the minimum value as the driving path of the logistics trolley.

[0014] Preferably, the original path image of the logistics trolley is determined according to the driving path, comprising:

[0015] Predefining a road node image library, the road node image library comprising a plurality of road nodes, each road node being provided with a corresponding node image;

[0016] Analyzing the driving path to determine the road nodes passed through in the driving path;

[0017] Filtering the road node image library according to the road nodes in the driving path to determine the node image corresponding to the road nodes in the driving path, and obtaining the original path image of the logistics trolley.

[0018] Preferably, the monitoring component comprises a camera, a laser radar, a millimeter wave radar and an inertial measurement unit;

[0019] The camera is used to take pictures when the logistics trolley is driving, and obtain video data of the driving path of the logistics trolley;

[0020] The laser radar and the millimeter wave radar are used to obtain radar data of the driving path of the logistics trolley when the logistics trolley is driving, the radar data comprising obstacle distance and obstacle speed;

[0021] The inertial measurement unit is used to obtain driving data of the logistics trolley when the logistics trolley is driving, the driving data comprising the driving speed of the logistics trolley.

[0022] Preferably, based on the original path image, the video data is analyzed to determine whether there is an obstacle, comprising:

[0023] Positioning the logistics trolley to obtain the real-time positioning of the logistics trolley, and determining the time corresponding to the coincidence of the real-time positioning and each road node in the driving path;

[0024] According to the time, the video data is frame extracted to obtain the real-time path image corresponding to each road node in the driving path;

[0025] target detection is performed on the real-time path image and the original path image corresponding to the road node, and a plurality of targets in the real-time path image and the original path image are determined;

[0026] The plurality of targets in the real-time path image are compared with the plurality of targets in the original path image one by one, and if there is a target that does not match, it is determined that there is an obstacle;

[0027] If the plurality of targets in the real-time path image and the plurality of targets in the original path image can be successfully matched, it is determined that there is no obstacle.

[0028] Preferably, the obstacle type is determined according to the video data and the radar data, comprising:

[0029] When the obstacle type is determined according to the video data, it comprises:

[0030] When it is determined that there is an obstacle, the target that does not match is marked to obtain a marked target, and the position of the marked target is determined; the position of the marked target at the current moment is compared with the position of the marked target at the last moment to determine whether the marked target at the current moment and the marked target at the last moment are at the same position, if they are at the same position, it is determined that the marked target is a static obstacle; if they are not at the same position, it is determined that the marked target is a dynamic obstacle;

[0031] When the obstacle type is determined according to the radar data, it comprises:

[0032] The radar data includes obstacle distance and obstacle speed; the obstacle speed at the current moment and the obstacle speed at the last moment are obtained, the obstacle acceleration is determined based on the obstacle speed at the current moment and the obstacle speed at the last moment, and it is judged whether the obstacle speed at the current moment, the obstacle speed at the last moment and the obstacle acceleration are all 0, if they are all 0, it is determined that the obstacle is a static obstacle; if they are not all 0, it is determined that the obstacle is a dynamic obstacle;

[0033] If the obstacle type is determined to be a static obstacle according to the video data, and the obstacle type is determined to be a static obstacle according to the radar data, it is determined that the obstacle type is a static obstacle;

[0034] Otherwise, it is determined that the obstacle type is a dynamic obstacle.

[0035] Preferably, the collision risk value of the logistics trolley and the obstacle is determined according to the radar data, the driving data and the obstacle type, comprising:

[0036] The safety distance ratio between the obstacle and the logistics trolley is determined according to the obstacle distance and the driving speed of the logistics trolley;

[0037] determine a safety time ratio between the obstacle and the logistics cart according to the obstacle distance, the obstacle speed and the driving speed of the logistics cart;

[0038] determine a collision risk value of the logistics cart and the obstacle according to the safety distance ratio, the safety time ratio and the obstacle type;

[0039] wherein the collision risk value is determined according to the following formula:

[0040] ;

[0041] R represents the collision risk value, D represents the obstacle distance, Vc represents the driving speed of the logistics cart, Tr represents the braking response time of the logistics cart, Vo represents the obstacle speed, C represents the risk amount corresponding to the obstacle type, and a, β, γ are weight coefficients.

[0042] Preferably, a collision risk level is determined according to the collision risk value, comprising:

[0043] a first collision risk value and a second collision risk value are preset, the first collision risk value being smaller than the second collision risk value;

[0044] a collision risk level is determined according to the relationship between the collision risk value and the first collision risk value and the second collision risk value;

[0045] if the collision risk value is smaller than the first collision risk value, the collision risk level is determined as a low risk level;

[0046] if the collision risk value is greater than or equal to the first collision risk value and smaller than the second collision risk value, the collision risk level is determined as a medium risk level;

[0047] if the collision risk value is greater than or equal to the second collision risk value, the collision risk level is determined as a high risk level.

[0048] Preferably, an obstacle avoidance instruction is set according to the collision risk level, comprising:

[0049] if the collision risk level is a low risk level, the logistics cart is set to drive along the driving path and monitor the environment in real time;

[0050] if the collision risk level is a medium risk level, the logistics cart is set to drive along the driving path, the driving speed is reduced by 30%-50%, and the lane offset of the logistics cart is set to a lane without obstacles;

[0051] if the collision risk level is a high risk level, the logistics cart is set to perform emergency braking.

[0052] The application further discloses an obstacle avoidance system for the automatic driving logistics trolley in the agricultural area.

[0053] A path image acquisition module is configured to acquire a plurality of destinations of the logistics trolley, formulate a driving path of the logistics trolley according to the destinations, and determine an original path image of the logistics trolley according to the driving path.

[0054] A driving monitoring module is configured to control the driving of the logistics trolley according to the driving path, and acquire video data, radar data and driving data acquired by the monitoring components of the logistics trolley.

[0055] An obstacle determination module is configured to analyze the video data based on the original path image, determine whether there is an obstacle, and if there is an obstacle, determine an obstacle type according to the video data and the radar data, wherein the obstacle type includes a dynamic obstacle and a static obstacle.

[0056] A collision risk determination module is configured to determine a collision risk value of the logistics trolley and the obstacle according to the radar data, the driving data and the obstacle type, and determine a collision risk level according to the collision risk value.

[0057] An obstacle avoidance control module is configured to set an obstacle avoidance instruction according to the collision risk level, and control the logistics trolley based on the obstacle avoidance instruction to realize obstacle avoidance.

[0058] Compared with the prior art, the application has the beneficial effects that the application combines vision, radar and inertial measurement data to improve the accuracy of obstacle detection and type identification. Through multi-parameter calculation of the collision risk value, the risk level is finely divided to avoid overreaction or insufficient reaction. According to the risk level, different control instructions are taken, and the driving efficiency and safety are considered. BRIEF DESCRIPTION OF DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute the embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of the provided drawings.

[0060] Figure 1 is a flowchart of an obstacle avoidance method for an automatic driving logistics trolley in an agricultural area;

[0061] Figure 2 is a function block diagram of an obstacle avoidance system for an automatic driving logistics trolley in an agricultural area. DETAILED DESCRIPTION

[0062] The technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work are within the protection scope of the present application.

[0063] As shown in the Figure 1 The present application provides an obstacle avoidance method for an automatic driving logistics trolley in an agricultural area, comprising:

[0064] S1, obtaining a plurality of destinations of a logistics trolley, formulating a driving path of the logistics trolley according to the destinations, and determining an original path image of the logistics trolley according to the driving path.

[0065] S2, controlling the logistics trolley to drive according to the driving path, and obtaining video data, radar data and driving data obtained by a monitoring component of the logistics trolley.

[0066] S3, based on the original path image, analyzing the video data to determine whether there is an obstacle, and if there is an obstacle, determining an obstacle type according to the video data and the radar data, wherein the obstacle type includes a dynamic obstacle and a static obstacle.

[0067] S4, determining a collision risk value of the logistics trolley and the obstacle according to the radar data, the driving data and the obstacle type, and determining a collision risk level according to the collision risk value.

[0068] S5, setting an obstacle avoidance instruction according to the collision risk level, and controlling the logistics trolley based on the obstacle avoidance instruction to realize obstacle avoidance.

[0069] The application can provide accurate reference for subsequent obstacle detection and obstacle avoidance operation by obtaining the destination and planning the driving path, and generating the original path image, thereby improving the accuracy and reliability of path planning. The video data, radar data and driving data are comprehensively analyzed to give full play to the advantages of each type of data. The video data can be used to visually identify the shape of the obstacle, the radar data can accurately measure the distance and speed, and the driving data provides real-time basis for dynamic adjustment, thereby realizing comprehensive environmental perception. Through joint analysis of video data and radar data, the obstacles are divided into dynamic and static types, and different obstacle avoidance strategies are adopted for different types of obstacles, thereby improving the pertinence and efficiency of obstacle avoidance. The collision risk value is calculated based on the radar data, driving data and obstacle type, and the risk level is further determined, so that the obstacle avoidance decision is more scientific and reasonable, and the misjudgment or delay caused by single threshold judgment is avoided. According to the collision risk level, specific obstacle avoidance instructions are set, and the logistics trolley is controlled to execute the obstacle avoidance action in real time, thereby ensuring the safety and stability of the logistics trolley in the complex agricultural environment, and improving the automation level of logistics transportation. The application effectively solves the obstacle avoidance problem of the agricultural automatic driving logistics trolley in the complex environment, significantly improves the safety, adaptability and operation efficiency of the logistics trolley, and provides important technical support for agricultural logistics automation.

[0070] The technical scheme of the present application will be described in detail:

[0071] S1, obtaining a plurality of destinations of a logistics trolley, planning a driving path of the logistics trolley according to the destinations, and determining an original path image of the logistics trolley according to the driving path.

[0072] In some embodiments of the present application, planning the driving path of the logistics trolley according to the destination comprises: obtaining current position information of the logistics trolley, generating a plurality of driving paths traversing all destinations by using a path planning algorithm; analyzing the plurality of driving paths to determine the driving distance of each driving path; comparing the driving distances to determine the minimum value of the driving distances, and setting the driving path corresponding to the minimum value as the driving path of the logistics trolley.

[0073] In the present embodiment, the current position information of the logistics trolley is acquired, which can be achieved by positioning technologies such as global positioning system (GPS), Beidou satellite navigation system, etc., to ensure accurate positioning of the logistics trolley in the rural area. Then, a path planning algorithm is used, such as A algorithm, Dijkstra algorithm, genetic algorithm, etc. classic algorithm, or path planning algorithm based on machine learning, to generate a number of driving paths that traverse all destinations. Taking A algorithm as an example, it combines the breadth-first search strategy of Dijkstra algorithm and the best-first search strategy of greedy algorithm, calculates the cost function of each node (including the actual cost from the starting point to the current node and the estimated cost from the current node to the target node), and selects the node with the minimum cost to expand, thereby finding the shortest path set from the starting point to all destinations. Then, the driving paths are analyzed, and the driving distance of each driving path is calculated, which can be accurately calculated through map data and path node information. Finally, the driving distances are compared, the minimum value of the driving distance is determined, and the driving path corresponding to the minimum value is set as the driving path of the logistics trolley, to ensure that the logistics trolley can travel in the shortest path during transportation, improve transportation efficiency, and reduce energy consumption.

[0074] In some embodiments of the present application, determining the original path image of the logistics trolley according to the driving path comprises: presetting a road node image library, the road node image library comprising a plurality of road nodes, each road node being provided with a corresponding node image; analyzing the driving path to determine the road nodes passed through in the driving path; screening the road node image library according to the road nodes in the driving path to determine the node images corresponding to the road nodes in the driving path, and obtaining the original path image of the logistics trolley.

[0075] In the present embodiment, a road node image library is established in advance, which covers various types of road nodes in the rural area, and each road node is equipped with a corresponding node image. These images can be obtained by field shooting, satellite image analysis or map data conversion, etc. The determined driving path is analyzed to determine the road nodes passed through in the driving path. According to the road nodes in the driving path, the road node image library is screened to find the node images corresponding to the road nodes in the driving path. These node images are combined in order to obtain the original path image of the logistics trolley. For example, if the driving path passes through three road nodes A, B and C, the corresponding images A', B' and C' are found in the road node image library and arranged in order to form the original path image, which provides a basic reference image for subsequent obstacle detection.

[0076] S2, controlling the logistics trolley to drive according to the driving path, and acquiring the video data, radar data and driving data obtained by the monitoring components of the logistics trolley.

[0077] In some embodiments of the present application, the monitoring components include a camera, a laser radar, a millimeter wave radar, and an inertial measurement unit; the camera is used to take pictures when the logistics trolley is driving, to obtain video data of the driving path of the logistics trolley; the laser radar and the millimeter wave radar are used to obtain radar data of the driving path of the logistics trolley when the logistics trolley is driving, the radar data including obstacle distance and obstacle speed; the inertial measurement unit is used to obtain driving data of the logistics trolley when the logistics trolley is driving, the driving data including driving speed of the logistics trolley.

[0078] In the present embodiment, the logistics trolley relies on multiple monitoring components to obtain key data during driving. The camera is installed at a suitable position of the logistics trolley, such as the front, the rear, the sides of the body, etc., to ensure that it can comprehensively capture video data of the driving path of the logistics trolley. These video data can be used to visually observe the situation on the driving path and provide visual information for obstacle detection and identification. The laser radar and the millimeter wave radar work together, the laser radar obtains high-precision information such as obstacle distance, angle, shape by emitting laser beams and receiving reflected signals, and the millimeter wave radar detects obstacles using millimeter wave band electromagnetic waves, which can work stably in bad weather conditions and obtain information such as obstacle speed. The data of the two are complementary to each other and together provide radar data of the driving path for the logistics trolley. The inertial measurement unit (IMU) is installed on the main structure of the logistics trolley to obtain driving data of the logistics trolley in real time, including driving speed, acceleration, attitude, etc. These data are crucial for evaluating the motion state and collision risk of the logistics trolley. Through the cooperative work of these monitoring components, the logistics trolley can comprehensively and accurately obtain various data during driving, providing strong support for subsequent obstacle avoidance decisions.

[0079] S3, based on the original path image, analyzing the video data to determine whether there is an obstacle, if there is an obstacle, determining the obstacle type according to the video data and the radar data, the obstacle type including dynamic obstacles and static obstacles.

[0080] In some embodiments of the present application, based on the original path image, the video data is analyzed to determine whether there is an obstacle, including: positioning the logistics trolley, obtaining the real-time positioning of the logistics trolley, and determining the time when the real-time positioning coincides with each road node in the driving path; according to the time, the video data is frame extracted to obtain the real-time path image corresponding to each road node in the driving path; target detection is performed on the real-time path image and the original path image of the corresponding road node to determine a plurality of targets in the real-time path image and the original path image; the plurality of targets in the real-time path image and the plurality of targets in the original path image are compared one by one, and if there is a target that does not match, it is determined that there is an obstacle; if the plurality of targets in the real-time path image and the plurality of targets in the original path image can all be successfully matched, it is determined that there is no obstacle.

[0081] In the present embodiment, the logistics trolley is accurately positioned by using satellite positioning system, base station positioning or sensor fusion-based positioning technology to obtain real-time positioning information of the logistics trolley. The time when the real-time positioning coincides with each road node in the driving path is determined, which can be achieved by matching the real-time positioning information with the pre-stored road node position information. According to these times, the video data is frame extracted to obtain the real-time path image corresponding to each road node in the driving path, ensuring that the obtained image can accurately reflect the environmental conditions of the logistics trolley at the key positions. Then, target detection is performed on the real-time path image and the original path image of the corresponding road node, and a plurality of targets in the image can be identified by using a target detection algorithm based on deep learning, such as Faster R-CNN, YOLO series, etc. The plurality of targets in the real-time path image and the plurality of targets in the original path image are compared one by one, and if there is a target that does not match, i.e. a target appears in the real-time path image that does not exist in the original path image, or a target in the original path image disappears in the real-time path image, it is determined that there is an obstacle; if the plurality of targets in the real-time path image and the plurality of targets in the original path image can all be successfully matched, it means that the environment on the driving path is consistent with the pre-set original path image, and it is determined that there is no obstacle.

[0082] In some embodiments of the present application, the type of obstacle is determined according to the video data and the radar data, including: when determining the type of obstacle according to the video data, the target that does not match is marked to obtain a marked target, and the position of the marked target is determined; the position of the marked target at the current time is compared with the position of the marked target at the last time to determine whether the marked target at the current time and the marked target at the last time are at the same position, if they are at the same position, it is determined that the marked target is a static obstacle; if they are not at the same position, it is determined that the marked target is a dynamic obstacle;

[0083] According to the radar data, the obstacle type is determined, including: the radar data includes obstacle distance and obstacle speed; the current time obstacle speed and the last time obstacle speed are obtained, the current time obstacle speed and the last time obstacle speed are used to determine the obstacle acceleration, and whether the current time obstacle speed, the last time obstacle speed and the obstacle acceleration are all 0 is judged; if all are 0, the obstacle is determined to be a static obstacle; if not all are 0, the obstacle is determined to be a dynamic obstacle.

[0084] If the obstacle type is determined to be a static obstacle according to the video data, and the obstacle type is determined to be a static obstacle according to the radar data, the obstacle type is determined to be a static obstacle; otherwise, the obstacle type is determined to be a dynamic obstacle.

[0085] In this embodiment, when determining the obstacle type according to the video data, the unmatching target is marked after determining the existence of the obstacle, the marked target is obtained, and the position of the marked target is determined through image analysis technology. The position of the current time marked target is compared with the position of the last time marked target, if they are in the same position, it is determined that the target does not move in this period of time, and the marked target is determined to be a static obstacle; if they are not in the same position, it is determined that the target moves, and the marked target is determined to be a dynamic obstacle. When determining the obstacle type according to the radar data, the current time obstacle speed and the last time obstacle speed are obtained, the obstacle acceleration is calculated based on the two speed values, and whether the current time obstacle speed, the last time obstacle speed and the obstacle acceleration are all 0 is judged; if all are 0, it is determined that the obstacle is in a static state, and the obstacle is determined to be a static obstacle; if not all are 0, it is determined that the obstacle has speed change and is in a motion state, and the obstacle is determined to be a dynamic obstacle. Finally, if the obstacle type is determined to be a static obstacle according to the video data and the radar data, the obstacle type is finally determined to be a static obstacle; otherwise, as long as one of the data is determined to be a dynamic obstacle, the obstacle type is determined to be a dynamic obstacle.

[0086] It can be understood that by combining video data and radar data, the obstacle is analyzed from two dimensions of visual information and physical motion characteristics respectively, which can effectively reduce the misjudgment caused by a single data source. For example, the video data can directly identify the existence of the obstacle and the change of its position, and the radar data provides accurate speed, distance and acceleration information, and the combination of the two significantly improves the reliability of static and dynamic obstacle classification. In a complex scene (such as light change, shielding or radar signal noise), a single sensor may fail or produce errors. The present scheme uses the complementarity of video and radar data through multi-source data fusion, reduces the influence of environmental interference on obstacle detection, and improves the stability and adaptability of the system in various practical application scenarios.

[0087] By comparing the time series of the target position in the video data and jointly analyzing the speed and acceleration in the radar data, the scheme can not only distinguish between static and dynamic obstacles, but also further capture the motion trend (such as acceleration, deceleration, or uniform speed) of the dynamic obstacles. This refined analysis provides more rich input information for the subsequent decision module (such as path planning or obstacle avoidance strategy), which helps to improve the intelligent level of the overall system.

[0088] S4, determining a collision risk value of the logistics cart and the obstacle according to the radar data, driving data and obstacle type, and determining a collision risk level according to the collision risk value.

[0089] In some embodiments of the present application, determining the collision risk value of the logistics cart and the obstacle according to the radar data, driving data and obstacle type comprises: determining a safety distance ratio between the obstacle and the logistics cart according to the obstacle distance and the driving speed of the logistics cart; determining a safety time ratio between the obstacle and the logistics cart according to the obstacle distance, the obstacle speed and the driving speed of the logistics cart; determining the collision risk value of the logistics cart and the obstacle according to the safety distance ratio, the safety time ratio and the obstacle type;

[0090] wherein the collision risk value is determined according to the following formula:

[0091] ;

[0092] R represents the collision risk value, D represents the obstacle distance, Vc represents the driving speed of the logistics cart, Tr represents the braking response time of the logistics cart, Vo represents the obstacle speed, C represents the risk amount corresponding to the obstacle type, and a, β, γ are weight coefficients.

[0093] In this embodiment, the safety distance ratio between the obstacle and the logistics cart is calculated according to the obstacle distance in the radar data and the driving speed of the logistics cart obtained by the inertial measurement unit, which reflects the relative relationship between the current obstacle distance and the safety distance. At the same time, the safety time ratio between the obstacle and the logistics cart is calculated according to the obstacle distance, the obstacle speed and the driving speed of the logistics cart, which reflects the comparison between the time required for the logistics cart to meet the obstacle at the current speed and the safety time. Then, the collision risk value of the logistics cart and the obstacle is determined by comprehensively considering the safety distance ratio, the safety time ratio and the obstacle type.

[0094] In this embodiment, a, β, γ can be set through subsequent experimental results, such as a, β, γ can be set to 0.3, 0.5, 0.2.

[0095] In some embodiments of the present application, determining the collision risk level according to the collision risk value comprises: presetting a first collision risk value and a second collision risk value, the first collision risk value being less than the second collision risk value; determining the collision risk level according to the relationship between the collision risk value and the first collision risk value and the second collision risk value; if the collision risk value is less than the first collision risk value, determining the collision risk level as a low risk level; if the collision risk value is greater than or equal to the first collision risk value and less than the second collision risk value, determining the collision risk level as a medium risk level; and if the collision risk value is greater than or equal to the second collision risk value, determining the collision risk level as a high risk level.

[0096] S5, setting an obstacle avoidance instruction according to the collision risk level, and controlling the logistics trolley based on the obstacle avoidance instruction to realize obstacle avoidance.

[0097] In some embodiments of the present application, setting the obstacle avoidance instruction according to the collision risk level comprises: if the collision risk level is a low risk level, setting the logistics trolley to travel along the travel path and monitor the environment in real time; if the collision risk level is a medium risk level, setting the logistics trolley to travel along the travel path and reduce the travel speed by 30%-50%, and offset the lane of the logistics trolley to a lane without obstacles; and if the collision risk level is a high risk level, setting the logistics trolley to perform emergency braking.

[0098] In the present embodiment, if the collision risk level is a low risk level, it indicates that the current travel environment of the logistics trolley is relatively safe, and the logistics trolley is set to travel normally along the travel path and continuously monitor the environment in real time to timely discover potential risk changes. If the collision risk level is a medium risk level, it indicates that there is a certain collision risk, and at this time the logistics trolley is set to travel along the travel path, but the travel speed is reduced by 30%-50% to reduce the impact force when colliding. At the same time, the lane of the logistics trolley is offset to a lane without obstacles by using the path planning algorithm and sensor data to avoid potential collision threats. If the collision risk level is a high risk level, it means that the collision risk is extremely high, and at this time the logistics trolley is set to perform emergency braking, so that the logistics trolley stops as soon as possible by starting the emergency braking system of the vehicle, such as quickly applying braking force, cutting off power output, etc., to avoid collision accidents.

[0099] As shown in Figure 2 The present application also discloses an obstacle avoidance system for an automatic logistics trolley in an agricultural area, which is used for the above-mentioned obstacle avoidance method for an automatic logistics trolley in an agricultural area, and comprises:

[0100] A path image acquisition module is configured to acquire a plurality of destinations of the logistics trolley, formulate a driving path of the logistics trolley according to the destinations, and determine an original path image of the logistics trolley according to the driving path;

[0101] A driving monitoring module is configured to control the logistics trolley to drive according to the driving path, and acquire video data, radar data and driving data acquired by the monitoring components of the logistics trolley;

[0102] An obstacle determination module is configured to analyze the video data based on the original path image, determine whether there is an obstacle, and if there is an obstacle, determine an obstacle type according to the video data and the radar data, the obstacle type including a dynamic obstacle and a static obstacle;

[0103] A collision risk determination module is configured to determine a collision risk value of the logistics trolley and the obstacle according to the radar data, the driving data and the obstacle type, and determine a collision risk level according to the collision risk value;

[0104] An obstacle avoidance control module is configured to set an obstacle avoidance instruction according to the collision risk level, and control the logistics trolley to realize obstacle avoidance based on the obstacle avoidance instruction.

[0105] The path image acquisition module combines the destinations of the logistics trolley with the driving path to generate the original path image. This design not only improves the accuracy of path planning, but also provides a visual reference for subsequent obstacle analysis, thereby optimizing the overall navigation efficiency. The driving monitoring module integrates video data, radar data and driving data to achieve comprehensive monitoring of the running state of the logistics trolley. The collaborative use of multi-source data can effectively improve the environmental perception ability and ensure the stable operation of the vehicle in complex agricultural environments. The obstacle determination module accurately identifies obstacles and classifies them into dynamic or static types based on the original path image and video and radar data. This hierarchical determination mechanism improves the accuracy of obstacle detection and reduces the misjudgment rate, laying a reliable foundation for subsequent obstacle avoidance decisions. The collision risk determination module quantifies the collision risk value and classifies it by integrating radar data, driving data and obstacle type. This systematic risk assessment method can dynamically adjust according to actual conditions to ensure the safety of the logistics trolley in different scenarios. The obstacle avoidance control module generates an obstacle avoidance instruction according to the collision risk level and controls the logistics trolley to complete the obstacle avoidance operation in real time. The intelligent design of this module significantly improves the autonomous decision-making ability of the vehicle and enhances its adaptability and reliability in complex agricultural environments.

[0106] To further illustrate the technical solutions of the present application, the actual scene is simulated and described.

[0107] Simulation scene building:

[0108] A logistics transportation scenario in a large farm is set. The starting point of the logistics vehicle is located in the warehouse center of the farm, where various agricultural materials such as seeds, fertilizers, pesticides, and harvested agricultural products are stored. Multiple destinations are distributed in different areas of the farm, including various planting areas, breeding areas, and agricultural product processing points and sales points around the farm. For example, planting area A needs to transport a batch of new seeds and fertilizers for the upcoming sowing operation; breeding area B urgently needs to replenish feed; agricultural product processing point C is waiting to receive fresh vegetables for processing; and sales point D needs to distribute packaged agricultural products to meet market demand.

[0109] In this scenario, obstacles may be widely distributed. There may be temporary farm tools such as hoes and plows placed by farmers on the field paths, which are static obstacles; in the planting area, irrigation equipment may occupy part of the road space when working, also constituting static obstacles. In terms of dynamic obstacles, farm animals such as cows and sheep may walk randomly on the road; farmers driving farm machinery such as tractors and harvesters may also appear on the logistics vehicle's driving path at different times; in addition, there may be workers working in the farm, which increases the complexity and danger of logistics transportation.

[0110] Detailed step demonstration:

[0111] Path planning: After the logistics vehicle starts, it first obtains its current position information in the warehouse center through the built-in GPS module. Then, the vehicle-mounted computer uses the A* path planning algorithm to generate a number of driving paths that traverse each destination (planting area A, breeding area B, agricultural product processing point C, and sales point D) in combination with the electronic map data of the farm. These paths take into account the actual conditions of the road, such as the width, flatness, and whether the logistics vehicle is allowed to pass, etc. The distance of each driving path is calculated through the coordinate information in the map data and the distance formula between path nodes. After comparison, the shortest driving path is selected as the driving path of the logistics vehicle, which will make the logistics vehicle consume the least time and energy when completing the transportation task.

[0112] Data acquisition: During the driving of the logistics trolley along the planned path, the high-definition camera installed on the trolley head captures video data of the driving path at a speed of 30 frames per second. These video data are transmitted in real time to the image storage module of the on-board computer, providing intuitive image information for subsequent obstacle detection. The laser radar emits laser beams at a frequency of 10 times per second and receives the reflected signals, thereby obtaining high-precision radar data such as obstacle distance and angle. The millimeter wave radar continuously monitors the speed of the obstacle and updates the data every 0.1 second. The inertial measurement unit monitors the driving speed of the logistics trolley in real time and synchronously transmits the data to the data analysis module of the on-board computer, ensuring accurate control of the motion state of the logistics trolley.

[0113] Obstacle detection and identification: During the driving of the logistics trolley, the GPS positioning information and the pre-set road node position information are used to determine the time when each road node in the real-time positioning and driving path coincides. For example, when the logistics trolley drives to the road node 100 meters away from the entrance of planting area A, the system records the time at that moment. According to these time points, the video data captured by the camera are frame-extracted to obtain the real-time path image corresponding to the road node. Advanced YOLOv8 target detection algorithm is used to detect the targets in the real-time path image and the original path image of the corresponding road node, such as road signs, buildings, obstacles, etc. When the logistics trolley drives to a certain road node, an object appears in the real-time path image that does not exist in the original path image, which is determined to be a temporarily placed farm tool through comparison, i.e., an obstacle is determined to exist. After determining the existence of the obstacle, the unmatched target (farm tool) is marked, and its position is determined through image analysis algorithm. The position of the marked target at the current time is compared with the position of the marked target at the last time, and it is found that the position does not change. According to the video data, it is determined that the obstacle is a static obstacle. At the same time, the speed of the obstacle at the current time and the speed of the obstacle at the last time are obtained from the radar data, and the acceleration of the obstacle is calculated to be 0, and the speed of the obstacle at the current time and the speed of the obstacle at the last time are both 0. According to the radar data, it is also determined that the obstacle is a static obstacle, and finally the type of the obstacle is determined to be a static obstacle.

[0114] Collision risk assessment: According to the obstacle distance obtained by the laser radar and the driving speed of the logistics trolley obtained by the inertial measurement unit, the safety distance ratio between the obstacle and the logistics trolley is calculated. Assuming that the obstacle distance is 5 meters, the driving speed of the logistics trolley is 10 meters per second, and the brake response time is 0.5 seconds, the safety distance is the distance traveled by the logistics trolley within the brake response time, i.e. 10 x 0.5 = 5 meters, and the safety distance ratio is 5 ÷ 5 = 1. At the same time, according to the obstacle distance, the obstacle speed (0, because it is a static obstacle) and the driving speed of the logistics trolley, the safety time ratio between the obstacle and the logistics trolley is calculated, assuming that the logistics trolley needs 0.5 seconds to travel at the current speed to the obstacle, and the brake response time is 0.5 seconds, the safety time ratio is 0.5 ÷ 0.5 = 1. Knowing that the obstacle type is a static obstacle, the corresponding risk amount C is set to 0.5, the weight coefficients α = 0.4, β = 0.3, and γ = 0.3, according to the collision risk value calculation formula, R = 0.4 x 1 + 0.3 x 1 + 0.3 x 0.5 = 0.4 + 0.3 + 0.15 = 0.85. The first collision risk value is set to 0.5 and the second collision risk value is set to 1.0. Since 0.85 is greater than 0.5 and less than 1.0, the collision risk level is determined to be the medium risk level.

[0115] Obstacle avoidance execution: Since the collision risk level is the medium risk level, after the control system of the logistics trolley receives the obstacle avoidance instruction, the driving speed is first reduced by 40% from the original 10 meters per second to 6 meters per second. At the same time, the path planning algorithm in the vehicle-mounted computer and the sensor data are used to analyze the surrounding environment in real time, and it is found that there is a temporary lane with no obstacles beside it, so the logistics trolley is controlled to turn and drive into the lane, continue to drive according to the driving path, and continuously monitor the surrounding environment to ensure safe passage through the area.

[0116] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems or computer program products. Accordingly, the application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0117] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart or flowsheet block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0118] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart or flowsheet block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0119] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart or flowsheet block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0120] Finally, it should be noted that the above-mentioned embodiments are merely intended for describing the technical solutions of the present application, but not for limiting it. Although the present application is described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. An obstacle avoidance method for autonomous logistics vehicles in agricultural areas, characterized in that, include: Obtain several destinations of the logistics vehicle, formulate the driving route of the logistics vehicle based on the destinations, and determine the original path image of the logistics vehicle based on the driving route. The logistics vehicle is controlled to drive according to the driving path, and video data, radar data and driving data acquired by the logistics vehicle monitoring component are obtained. Based on the original path image, the video data is analyzed to determine whether there are obstacles. If there are obstacles, the type of obstacle is determined based on the video data and radar data. The type of obstacle includes dynamic obstacles and static obstacles. Based on the radar data, driving data and obstacle type, the collision risk value between the logistics vehicle and the obstacle is determined, and the collision risk level is determined based on the collision risk value. Obstacle avoidance commands are set according to the collision risk level, and the logistics vehicle is controlled based on the obstacle avoidance commands to achieve obstacle avoidance.

2. The obstacle avoidance method for an autonomous logistics vehicle in agricultural areas according to claim 1, characterized in that, The route for the logistics vehicle is determined based on the destination, including: Obtain the current location information of the logistics vehicle and use a path planning algorithm to generate several driving routes that traverse all destinations; Analyze several driving routes to determine the driving distance of each route; The driving distances are compared to determine the minimum value, and the driving path corresponding to the minimum value is set as the driving path of the logistics vehicle.

3. The obstacle avoidance method for an autonomous logistics vehicle in agricultural areas according to claim 1, characterized in that, Determining the original path image of the logistics vehicle based on the driving path includes: A road node image library is pre-set, which includes a number of road nodes, and each road node is provided with a corresponding node image. The driving path is analyzed to determine the road nodes traversed along the driving path; The road node image library is filtered based on the road nodes in the driving path to determine the node images corresponding to the road nodes in the driving path, thus obtaining the original path image of the logistics vehicle.

4. The obstacle avoidance method for an autonomous logistics vehicle in agricultural areas according to claim 1, characterized in that, The monitoring components include a camera, lidar, millimeter-wave radar, and inertial measurement unit; The camera is used to take pictures while the logistics vehicle is moving, and to acquire video data of the logistics vehicle's driving path; The lidar and millimeter-wave radar are used to acquire radar data of the logistics vehicle's travel path when the logistics vehicle is in motion. The radar data includes obstacle distance and obstacle speed. The inertial measurement unit is used to acquire the driving data of the logistics vehicle when it is in motion, and the driving data includes the speed of the logistics vehicle.

5. The obstacle avoidance method for an autonomous logistics vehicle in agricultural areas according to claim 3, characterized in that, Based on the original path image, the video data is analyzed to determine whether obstacles exist, including: The logistics vehicle is located, its real-time location is obtained, and the time when the real-time location coincides with each road node in the driving path is determined. Based on the stated time, frames are extracted from the video data to obtain a real-time path image corresponding to each road node in the driving path. Target detection is performed on the real-time path image and the original path image of the corresponding road node to identify several targets in the real-time path image and the original path image; The targets in the real-time path image are compared one by one with the targets in the original path image. If there are any mismatched targets, then it is determined that there is an obstacle. If several targets in the real-time path image can be successfully matched with several targets in the original path image, then it is determined that there are no obstacles.

6. The obstacle avoidance method for an autonomous logistics vehicle in agricultural areas according to claim 5, characterized in that, The obstacle type is determined based on the video data and radar data, including: When determining the obstacle type based on the video data, the following are included: When an obstacle is identified, mismatched targets are marked to obtain marked targets, and the position of the marked targets is determined. The position of the marked target at the current moment is compared with the position of the marked target at the previous moment to determine whether the marked target at the current moment is in the same position as the marked target at the previous moment. If they are in the same position, the marked target is determined to be a static obstacle; if they are not in the same position, the marked target is determined to be a dynamic obstacle. When determining the type of obstacle based on the radar data, the following are included: The radar data includes obstacle distance and obstacle speed; the obstacle speed at the current moment and the obstacle speed at the previous moment are obtained, the obstacle acceleration is determined based on the obstacle speed at the current moment and the obstacle speed at the previous moment, and it is determined whether the obstacle speed at the current moment, the obstacle speed at the previous moment, and the obstacle acceleration are all 0. If they are all 0, the obstacle is determined to be a static obstacle; if they are not all 0, the obstacle is determined to be a dynamic obstacle. If the obstacle type is determined to be a static obstacle based on the video data and the obstacle type is determined to be a static obstacle based on the radar data, then the obstacle type is determined to be a static obstacle. Otherwise, the obstacle type is determined to be a dynamic obstacle.

7. The obstacle avoidance method for an autonomous logistics vehicle in agricultural areas according to claim 4, characterized in that, The collision risk value between the logistics vehicle and the obstacle is determined based on the radar data, driving data, and obstacle type, including: The safe distance ratio between the obstacle and the logistics vehicle is determined based on the distance to the obstacle and the speed of the logistics vehicle. The safe time ratio between the obstacle and the logistics vehicle is determined based on the obstacle distance, obstacle speed, and logistics vehicle speed. The collision risk value between the logistics vehicle and the obstacle is determined based on the safe distance ratio, the safe time ratio, and the obstacle type. The collision risk value is determined according to the following formula: ； R represents the collision risk value, D represents the obstacle distance, Vc represents the speed of the logistics vehicle, Tr represents the braking response time of the logistics vehicle, Vo represents the obstacle speed, C represents the risk quantity corresponding to the obstacle type, and α, β, and γ are weighting coefficients.

8. The obstacle avoidance method for an autonomous logistics vehicle in agricultural areas according to claim 1, characterized in that, The collision risk level is determined based on the collision risk value, including: A first collision risk value and a second collision risk value are preset, wherein the first collision risk value is less than the second collision risk value; The collision risk level is determined based on the relationship between the collision risk value and the first and second collision risk values. If the collision risk value is less than the first collision risk value, then the collision risk level is determined to be a low risk level; If the collision risk value is greater than or equal to the first collision risk value and the collision risk value is less than the second collision risk value, then the collision risk level is determined to be a medium risk level. If the collision risk value is greater than or equal to the second collision risk value, then the collision risk level is determined to be a high-risk level.

9. The obstacle avoidance method for an autonomous logistics vehicle in agricultural areas according to claim 8, characterized in that, Based on the collision risk level, obstacle avoidance commands are set, including: If the collision risk level is low, the logistics vehicle is set to travel along the driving route and monitor the environment in real time. If the collision risk level is medium risk, the logistics vehicle is set to travel along the driving path and the driving speed is reduced by 30%-50%, while the logistics vehicle's lane is shifted to a lane without obstacles. If the collision risk level is high, then the logistics vehicle will be set to brake suddenly.

10. An obstacle avoidance system for an autonomous logistics vehicle in agricultural areas, used to apply the obstacle avoidance method for an autonomous logistics vehicle in agricultural areas as described in any one of claims 1-9, characterized in that, include: The path image acquisition module is configured to acquire several destinations of the logistics vehicle, determine the driving path of the logistics vehicle based on the destinations, and determine the original path image of the logistics vehicle based on the driving path. The driving monitoring module is configured to control the driving of the logistics vehicle according to the driving path and to acquire video data, radar data and driving data acquired by the logistics vehicle monitoring component; The obstacle determination module is configured to analyze the video data based on the original path image to determine whether there are obstacles. If there are obstacles, the obstacle type is determined based on the video data and radar data. The obstacle type includes dynamic obstacles and static obstacles. The collision risk determination module is configured to determine the collision risk value between the logistics vehicle and the obstacle based on the radar data, driving data and obstacle type, and to determine the collision risk level based on the collision risk value. The obstacle avoidance control module is configured to set obstacle avoidance commands based on the collision risk level, and control the logistics vehicle based on the obstacle avoidance commands to achieve obstacle avoidance.

Images