Dynamic obstacle avoidance method and system for mobile refueling device

Abstract

The application provides a dynamic obstacle avoidance method and system for a mobile storage and charging device, and relates to the technical field of obstacle avoidance control. The method comprises the following steps: determining the installed capacity of a target area for the operation of the mobile storage and charging device; constructing a dynamically updated hierarchical topological map and a Riemann manifold map; triggering an obstacle avoidance decision maker for a first obstacle target according to the movement of the mobile storage and charging device; performing branch master-slave deployment and obstacle avoidance decision under dynamic weight according to static obstacle targets or dynamic obstacle targets; determining a first obstacle avoidance strategy; the obstacle avoidance decision maker comprises a first obstacle avoidance branch constructed according to the hierarchical topological map and a second obstacle avoidance branch constructed based on the Riemann manifold map; and performing automatic obstacle avoidance control. The application solves the technical problem that the obstacle avoidance method in the prior art relies on a static map or a predetermined path planning, cannot adapt to sudden obstacles or environmental changes in real time, and affects the obstacle avoidance efficiency and safety of the mobile storage and charging device.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of obstacle avoidance control, in particular to a dynamic obstacle avoidance method and system for mobile storage and charging equipment. BACKGROUND

[0002] Mobile storage and charging equipment is widely used in various working environments, such as electric vehicle charging stations, battery exchange systems, warehouse logistics, etc. The main task of these mobile storage and charging equipment is to provide energy supplement, material transportation and other services in an automated manner. However, with the complication of application scenarios, dynamic obstacle avoidance has become an important technical problem faced by mobile storage and charging equipment. Many existing obstacle avoidance methods rely on static maps or predetermined path planning, which cannot adapt to sudden obstacles or environmental changes in real time. With the appearance of dynamic obstacles, mobile storage and charging equipment needs to update environmental perception and path planning strategies in real time to avoid being blocked by unpredictable obstacles. If there is a lack of real-time updating capability, it may lead to collision or path planning failure, thereby affecting the obstacle avoidance efficiency and safety of mobile storage and charging equipment. SUMMARY

[0003] The present application provides a dynamic obstacle avoidance method and system for mobile storage and charging equipment, aiming to solve the technical problem that the existing obstacle avoidance method relies on static maps or predetermined path planning, which cannot adapt to sudden obstacles or environmental changes in real time, affecting the obstacle avoidance efficiency and safety of mobile storage and charging equipment.

[0004] The first aspect of the present application provides a dynamic obstacle avoidance method for mobile storage and charging equipment, the method comprising: determining the installed capacity of a target area for the operation of the mobile storage and charging equipment; constructing a dynamically updated hierarchical topological map and a Riemann manifold map according to the installed capacity of the target area, wherein the information mode of each layer of the topological map is different, and the Riemann manifold map equivalent to the motion of obstacles as a Riemann curvature tensor; triggering an obstacle avoidance decision maker for a first obstacle target according to the motion of the mobile storage and charging equipment, and determining a first obstacle avoidance strategy according to the static obstacle target or the dynamic obstacle target under the dynamic weight of the branch master-slave deployment and obstacle avoidance decision of the obstacle avoidance decision maker, wherein the obstacle avoidance decision maker includes a first obstacle avoidance branch constructed according to the hierarchical topological map, and a second obstacle avoidance branch constructed based on the Riemann manifold map; and automatically controlling the walking mechanism of the mobile storage and charging equipment according to the first obstacle avoidance strategy.

[0005] In a second aspect, the application discloses a dynamic obstacle avoidance system for a mobile storage and charging device, which is used in the dynamic obstacle avoidance method for the mobile storage and charging device, and comprises: a regional installed capacity determination module configured to determine a regional installed capacity for a target region of the mobile storage and charging device; a Riemannian manifold graph construction module configured to construct a dynamic updated hierarchical topological map and a Riemannian manifold graph according to the regional installed capacity, wherein information modes of the topological maps of different layers are different, and the Riemannian manifold graph is used to equivalently convert obstacle movement into a Riemann curvature tensor; an obstacle avoidance strategy determination module configured to trigger an obstacle avoidance decision maker for a first obstacle target according to movement of the mobile storage and charging device, and determine a first obstacle avoidance strategy according to static obstacle targets or dynamic obstacle targets, branch master-slave deployment and obstacle avoidance decisions under dynamic weights of the obstacle avoidance decision maker, wherein the obstacle avoidance decision maker comprises a first obstacle avoidance branch constructed according to the hierarchical topological map and a second obstacle avoidance branch constructed according to the Riemannian manifold graph; and an automatic obstacle avoidance control module configured to automatically control a walking mechanism of the mobile storage and charging device according to the first obstacle avoidance strategy.

[0006] The one or more technical solutions provided in the application have at least the following beneficial effects:

[0007] By determining the regional installed capacity of the target region, the work range and equipment demand can be effectively evaluated, and appropriate path planning and obstacle avoidance strategies can be selected for the mobile storage and charging device according to actual size, obstacle density and other factors of the region; by constructing the dynamic updated hierarchical topological map, accurate path planning can be performed according to information of different levels, each topological map reflects environment information of different levels, thereby reducing the calculation amount and improving the efficiency and accuracy of path planning; by constructing the dynamic updated Riemannian manifold graph, obstacle movement is equivalently converted into a Riemann curvature tensor, so that the system can handle changes caused by dynamic obstacles, and the obstacle avoidance path can be adjusted in real time to ensure that the device can efficiently avoid obstacles; by triggering the obstacle avoidance decision maker, obstacles can be identified in real time during actual movement of the mobile storage and charging device, and obstacle avoidance strategies can be adjusted according to different types of obstacles; the dynamic weight setting ensures that the main branch and the auxiliary branch can be flexibly selected according to actual conditions in the case of static obstacles and dynamic obstacles, and this dynamic adjustment avoids excessively conservative or excessively aggressive obstacle avoidance strategies, thereby improving the obstacle avoidance efficiency and safety; after the first obstacle avoidance strategy is determined, the walking mechanism of the mobile storage and charging device is adjusted in real time to ensure that the device can travel along an optimal path while avoiding collision with obstacles, and manual intervention is greatly reduced through automatic control, thereby improving work efficiency and safety.

[0008] The above description is only a summary of the technical solutions of the present application. In order to enable the technical means of the present application to be more clearly understood, and to be implemented according to the content of the description, and in order to enable the above and other purposes, characteristics and advantages of the present application to be more apparent and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1 A flowchart of a dynamic obstacle avoidance method for a mobile storage and charging device is provided for the embodiments of the present application.

[0010] Figure 2 A schematic diagram of the structure of a dynamic obstacle avoidance system for a mobile storage and charging device is provided for the embodiments of the present application.

[0011] Explanation of reference signs: regional installation scale determination module 10, Riemann manifold graph construction module 20, obstacle avoidance strategy determination module 30, automatic obstacle avoidance control module 40. DETAILED DESCRIPTION

[0012] The embodiments of the present application provide a dynamic obstacle avoidance method and system for a mobile storage and charging device, which solves the technical problem that the obstacle avoidance method of the prior art relies on a static map or a predetermined path planning, and cannot adapt to sudden obstacles or environmental changes in real time, affecting the obstacle avoidance efficiency and safety of the mobile storage and charging device.

[0013] After introducing the basic principles of the present application, the various non-limiting embodiments of the present application will be specifically described below in conjunction with the drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0014] Embodiment one, as shown in the drawings, the embodiments of the present application provide a dynamic obstacle avoidance method for a mobile storage and charging device, the method comprising: Figure 1 Determining the regional installation scale of the target region of the mobile storage and charging device operation.

[0015] The target region of the mobile storage and charging device operation is clear, and the target region is a specific physical environment, such as an industrial park, etc. The regional installation scale refers to the size, operation complexity and obstacle distribution of the target region in the target region of the mobile storage and charging device operation, as well as the number, type and distribution of the deployed devices, which provides a basis for subsequent obstacle avoidance decision-making, and ensures that the device can complete the operation task efficiently and safely in a complex environment.

[0016] According to the regional installation scale, a dynamically updated hierarchical topological map and a Riemann manifold graph are constructed, wherein the information mode of each layer of the topological map is different, and the Riemann manifold graph equivalent to the obstacle motion is equivalent to the Riemann curvature tensor.

[0017]

[0018] ​According to the installed capacity of the region, the basic map of the obstacle avoidance system is established, including a hierarchical topological map and a Riemannian manifold map, which respectively provide spatial structure information of static and dynamic obstacles for real-time obstacle avoidance decision-making of the device. Among them, the hierarchical topological map is used to express different levels of information in the environment, including a bottom layer topological map, a middle layer topological map, and a high layer topological map; the Riemannian manifold map is used to express the movement of dynamic obstacles and its influence. The hierarchical topological map and the Riemannian manifold map provide real-time and accurate spatial information for subsequent obstacle avoidance decision-making, supporting the device to effectively avoid obstacles in a dynamic environment.

[0019] With the movement of the mobile storage and charging device, the obstacle avoidance decision maker is triggered for the first obstacle target, and the branch master-slave deployment and obstacle avoidance decision of the obstacle avoidance decision maker under dynamic weight are made according to the static obstacle target or the dynamic obstacle target, to determine the first obstacle avoidance strategy, wherein the obstacle avoidance decision maker includes a first obstacle avoidance branch constructed according to the hierarchical topological map, and a second obstacle avoidance branch constructed based on the Riemannian manifold map.

[0020] In the operation process of the mobile storage and charging device, through sensors such as radar and camera, the front obstacles are constantly scanned, and when the sensor detects an obstacle and judges that there is a conflict between the obstacle and the path of the mobile storage and charging device, the first obstacle target is determined, which can be a dynamic obstacle (such as a moving vehicle) or a static obstacle (such as a fixed wall). After triggering the obstacle avoidance decision maker, the obstacle avoidance decision under dynamic weight is made, at this time, the obstacle avoidance decision maker adjusts the dynamic weight according to the output of the hierarchical topological map and the Riemannian manifold map, so as to select the appropriate obstacle avoidance strategy according to the type of obstacle (static or dynamic).

[0021] The branch master-slave deployment means that the main branch is selected according to the type of the current obstacle, and the auxiliary branch provides auxiliary decision for the main branch, for example, if the static obstacle is the main one, the first obstacle avoidance branch dominates the obstacle avoidance path, and the second obstacle avoidance branch assists; if the dynamic obstacle is the main one, the second obstacle avoidance branch dominates the obstacle avoidance path, and the first obstacle avoidance branch assists. According to the above dynamic weight and branch master-slave deployment, the first obstacle avoidance strategy is determined, which is a specific obstacle avoidance scheme on the path of the device, helping the device to avoid obstacles and continue to perform tasks.

[0022] According to the first obstacle avoidance strategy, the walking mechanism of the mobile storage and charging device is automatically controlled to avoid obstacles.

[0023] After the first obstacle avoidance strategy is generated, the first obstacle avoidance strategy is converted into a control signal, including changing the movement speed and turning angle of the device, to instruct the walking mechanism of the mobile storage and charging device how to adjust the movement mode, so as to ensure that the mobile storage and charging device can avoid obstacles and successfully complete the task.

[0024] Further, a dynamically updated hierarchical topological map is constructed, including:

[0025] For the target area, a three-dimensional coordinate space is constructed, in which the bottom-level topological map is determined based on grid and point cloud geometric space segmentation, the middle-level topological map is determined based on semantic segmentation network and obstacle type labeling, and the high-level topological map is determined based on passable area as node and semantic risk as edge weight. The bottom-level topological map, the middle-level topological map and the high-level topological map are mapped according to the space phase as the hierarchical topological map.

[0026] A three-dimensional coordinate space is constructed for the target area, which is a spatial representation of the entire environment. It is usually constructed based on spatial data obtained by laser radar, stereo vision or depth camera, etc. Through the three-dimensional coordinate space, the position of each point in the target area can be accurately represented as a three-dimensional coordinate (x, y, z).

[0027] Griding is the division of a three-dimensional coordinate space into discrete grid cells, each grid cell representing a certain area of spatial information. The size of each grid cell is adjusted according to actual needs. Generally, the smaller the grid, the higher the accuracy of the map, but the computational load will also increase. Point cloud refers to three-dimensional spatial data obtained by laser scanning and other devices, containing a large number of spatial points. Through point cloud segmentation, the point cloud data in space can be separated from the obstacle-free area according to the distribution of obstacles. Point cloud segmentation helps to determine the location of obstacles and their geometric shape. Through gridding and point cloud segmentation, the bottom-level topological map is determined. The bottom-level topological map is used to represent the basic geometric information of the target area, including the location, size and shape of obstacles.

[0028] Semantic segmentation network is a deep learning-based image processing method that assigns each pixel in an image to a class. In a three-dimensional coordinate space, semantic segmentation network is used to classify different types of obstacles, such as walls, devices, and people, which can be identified through semantic segmentation network. By introducing semantic understanding, it can identify which obstacles are passable, such as open spaces or spaces to pass through, and which are completely impassable, such as walls, which avoids overly conservative path planning, such as always avoiding all obstacles, and instead makes reasonable adjustments based on the passability of obstacles. Obstacle type labeling is performed because different types of obstacles will affect the device's obstacle avoidance strategy, for example, static obstacles (such as walls) and dynamic obstacles (such as pedestrians or vehicles) have different requirements for path planning. Through semantic segmentation and obstacle type labeling, the middle-level topological map is constructed. The middle-level topological map adds semantic information of obstacles to the bottom-level topological map, which can express the category, type and risk level of obstacles in the area, etc.

[0029] On the basis of the middle-layer topological map, a passable region in the environment is further abstracted, the passable region being a space through which the device can safely pass, and the passable region is defined as a node in the high-layer topological map, for example, if a region contains static obstacles and there is enough space between the static obstacles for the device to pass through, the region is regarded as a passable region. Semantic risk refers to the influence of the type and number of obstacles and the complexity of the environment on the obstacle avoidance decision in path planning, for example, the risk of an area close to a moving obstacle is relatively high, and the risk of an area close to a static obstacle is relatively low, an edge in the high-layer topological map represents a path through which the device moves from one passable region to another passable region, and the weight of the edge reflects the risk and feasibility of the path, a high-risk path is given a larger edge weight, and a low-risk path has a smaller weight, so that the device will preferentially select a route with lower risk when selecting a path. In a three-dimensional coordinate space, a high-layer topological map is constructed based on the passable region and semantic risk information, the high-layer topological map expresses the relationship between different regions through the connection of nodes and edges, and provides a basis for preferentially selecting a path based on risk.

[0030] Phase mapping is a technology for fusing information at different levels, and here, the bottom-layer topological map, the middle-layer topological map and the high-layer topological map are integrated into a unified layered topological map through a spatial phase mapping technology, helping the device to make real-time path planning and obstacle avoidance decisions in a complex environment.

[0031] Further, a dynamically updated Riemannian manifold map is constructed, including:

[0032] A preset movement task of the mobile storage and charging device is acquired, and a first obstacle distribution of a task trajectory is determined, wherein the first obstacle distribution contains dynamic obstacle targets and static obstacle targets; for the first obstacle distribution, the motion of obstacles is equivalent to a Riemann curvature tensor, and a Riemannian manifold map is constructed.

[0033] The preset movement task is set in advance based on the working task of the mobile storage and charging device, and includes the starting point, the end point and the task trajectory of the mobile storage and charging device. According to the task trajectory, obstacle distribution information in the current region is acquired through an environment perception component such as a laser radar, a sensor, a camera and the like, wherein the motion trajectory of a dynamic obstacle target changes with time, for example, a vehicle in motion, a pedestrian and the like; the geometric shape and position of the dynamic obstacle target do not change with time.

[0034] Riemannian manifolds are manifolds with an inner product structure, which can be used to describe the geometric properties of space. In this process, the movement of obstacles is regarded as a change in the curvature of the manifold, which affects the feasibility and safety of the path. By equating the movement of obstacles to the local curvature change of the manifold, the dynamic influence of obstacles on the device path can be accurately described mathematically. Curvature is a basic geometric property of a manifold, which describes the degree of bending of the manifold. In the physical world, the movement of obstacles can be understood as an influence on the geometric shape of the local region. For example, when a dynamic obstacle approaches, the local curvature of the manifold increases, indicating that the feasibility of the path decreases; when the dynamic obstacle moves away, the local curvature decreases, and the feasibility of the path increases. The Riemann curvature tensor is a tool used in Riemannian geometry to describe the bending properties of a manifold, which indicates the degree of bending at different points of the manifold. In obstacle avoidance path planning, the Riemann curvature tensor can be used to describe the influence of obstacles on the geometric shape of the path at different times. By modeling the movement of obstacles as a change in curvature, it is possible to more accurately determine which areas are dangerous and which areas are safe to pass through. The change in curvature will affect the device's path selection, allowing it to avoid high-risk areas and choose a low-risk path.

[0035] Based on the influence of obstacle movement on the curvature of the manifold, a Riemannian manifold graph is constructed, which is a mathematical model used to describe the influence of obstacles on the geometric shape of space and the path. This graph will reflect the dynamic changes of obstacles over time and how these changes affect the geometric properties of the path. As the obstacles move, the Riemannian manifold graph is updated in real time. The construction of this dynamically updated Riemannian manifold graph allows the device to consider the dynamic influence of obstacles when planning a path, rather than relying solely on static geometric models. This allows the device to be more flexible in obstacle avoidance operations in a changing environment.

[0036] Furthermore, a dynamic curvature relationship is determined, and the Riemannian manifold graph is constructed based on the dynamic curvature relationship. The dynamic curvature relationship is the change in the curvature tensor components of the local manifold around the obstacle over time as the obstacle moves. When the obstacle approaches, the local curvature increases; when the obstacle moves away, the local curvature decreases. When the curvature exceeds the limit, an obstacle avoidance is triggered.

[0037] In Riemannian geometry, the curvature tensor describes the degree of bending of a manifold at a certain point, and the local curvature reflects the geometric changes of a certain region of space. The movement of obstacles will affect the local geometric properties of the manifold. Specifically, when the obstacle moves, it will change the curvature tensor components of the surrounding area, causing the local curvature of the area to change over time. The dynamic curvature relationship refers to the change in local curvature caused by the movement of the obstacle, which changes over time. Therefore, the curvature tensor components will dynamically adjust as the obstacle moves. By combining the change in local curvature with the dynamic movement of the obstacle, a Riemannian manifold graph is constructed. This Riemannian manifold graph will be dynamically updated as the obstacle moves, showing the influence of the obstacle on the geometric structure of space, and reflecting the results of the curvature change in the graph.

[0038] The local curvature of the manifold increases as the obstacle approaches, meaning that the current path becomes less suitable for passing through. For example, if the obstacle is approaching from the side, the travel space of the device will become narrower. Once the local curvature reaches a certain critical value, a danger is identified, triggering an obstacle avoidance mechanism to change the movement trajectory of the device and avoid the area. When the obstacle moves away, the local curvature decreases, indicating that the path of the device becomes more relaxed and can be safely passed through. At this time, the path becomes safer and normal travel can be resumed.

[0039] Further, after constructing the dynamically updated hierarchical topological map and Riemannian manifold graph, the following steps are included:

[0040] With the movement of the mobile storage and charging device, the detection components deployed at the front end of the device perform sensing and scanning to determine the target obstacle. It is determined whether the target obstacle belongs to the first obstacle distribution. If not, synchronization is performed in the hierarchical topological map and Riemannian manifold graph.

[0041] A series of detection components, such as lidar, ultrasonic sensor, camera, radar, etc., are installed at the front end of the mobile storage and charging device. As the mobile storage and charging device moves, it performs environmental scanning and real-time acquisition of environmental information around the device, identifying the position, shape, speed, and other characteristics of the obstacle. Through analysis of sensor data, different types of obstacles are identified and distinguished. For example, dynamic obstacles exhibit different movement patterns, while static obstacles remain stationary.

[0042] The first obstacle distribution is the distribution of obstacles that have been identified on the device's preset task path, including all static or dynamic obstacles that the device may encounter on the intended path. It is determined whether the type (static or dynamic) of the target obstacle matches the type of known obstacles in the first obstacle distribution. Specifically, the detected obstacle position is compared with the preset position of the obstacle in the first obstacle distribution. If the obstacle is located at the predetermined position of the task path and its attributes meet the expectations, such as a static obstacle or a known dynamic obstacle, it is considered that the target obstacle belongs to the first obstacle distribution.

[0043] If the target obstacle does not belong to the first obstacle distribution, it is added to the current environment model, including updating the hierarchical topological map and the Riemannian manifold map, so that the device can plan a path based on the latest environment information. This synchronization process enables the device to adapt to dynamically changing environments, ensuring that the device can maintain safe and smooth path planning even in the presence of sudden obstacles.

[0044] Further, before triggering the obstacle avoidance decision maker for the first obstacle target, the construction of the obstacle avoidance decision maker includes:

[0045] According to the hierarchical topological map, a first obstacle avoidance branch is constructed; according to the Riemannian manifold map, a second obstacle avoidance branch is constructed; the first and second obstacle avoidance branches are integrated as an obstacle avoidance decision maker and embedded in the intelligent central control of the mobile storage and charging device.

[0046] The hierarchical topological map includes a bottom topological map, a middle topological map, and a high topological map. A candidate path is generated according to the hierarchical topological map, for example, a local feasible path is calculated using the bottom topological map, the path risk is weighted using the middle topological map combined with semantic information, and the global obstacle avoidance direction is selected using the high topological map. A first obstacle avoidance branch is generated according to the candidate path, emphasizing obstacle avoidance based on map-based rules and geometric constraints.

[0047] The Riemannian manifold map equivalent obstacle motion as Riemann curvature tensor, the device can predict the dynamic influence of obstacle on path through geometric curvature change, generate candidate path according to Riemannian manifold map, for example, calculate local curvature at current position and target path of device, when curvature exceeds preset threshold, determine need to avoid obstacle, predict safety of future path according to curvature change and obstacle motion, generate second obstacle avoidance branch according to candidate path, emphasize geometric perception and prediction of dynamic environment and obstacle motion.

[0048] The first and second obstacle avoidance branches are integrated as an obstacle avoidance decision maker, which can dynamically adjust the weights according to the type and risk of the obstacle: static obstacles prefer the first obstacle avoidance branch, and dynamic obstacles prefer the second obstacle avoidance branch. The obstacle avoidance decision maker is deployed in the intelligent central control of the mobile storage and charging device to directly control the device motion, realizing closed-loop obstacle avoidance.

[0049] Further, the obstacle avoidance decision maker is deployed with branch master-slave under dynamic weights and obstacle avoidance decision, to determine the first obstacle avoidance strategy, including:

[0050] determining a first obstacle target, wherein the first obstacle target is any one of the first obstacle distribution or the target obstacle; setting a first weight and a second weight, and according to the obstacle target type, performing dynamic weight deployment and obstacle avoidance decision for the first obstacle avoidance branch and the second obstacle avoidance branch, and determining a first obstacle avoidance strategy, wherein the first weight > the second weight.

[0051] The first obstacle target refers to any obstacle that needs to be avoided during operation. It can be a known obstacle in the first obstacle distribution or a new obstacle detected in real time during device movement.

[0052] The first weight and the second weight are set, and the first weight is greater than the second weight, which is used to dynamically adjust the dominance of the first obstacle avoidance branch and the second obstacle avoidance branch in the obstacle avoidance process. The weight value is set according to experience and historical data. According to the obstacle target type of the first obstacle target, the weight of the first obstacle avoidance branch and the second obstacle avoidance branch is dynamically adjusted. For example, when the first obstacle target is a static obstacle, the weight distribution is biased towards the first obstacle avoidance branch; when the first obstacle target is a dynamic obstacle, the weight distribution is biased towards the second obstacle avoidance branch. Through this dynamic weight adjustment, the obstacle avoidance decision maker can flexibly respond to complex and variable environments, ensuring that the device can effectively avoid obstacles of various types.

[0053] According to the dynamic weight adjustment result, the obstacle avoidance decision is made. The core is to decide which branch to use according to the weight size, i.e. the first branch or the second branch, and adjust the walking direction or speed of the device through obstacle avoidance action. The first obstacle avoidance strategy is a specific obstacle avoidance scheme on the device's travel path, which helps the device avoid obstacles and continue to perform tasks.

[0054] Further, if the first obstacle target is a static obstacle target, the first obstacle avoidance branch is identified with a first weight, the second obstacle avoidance branch is identified with a second weight, and a one-way interaction from the second obstacle avoidance branch to the first obstacle avoidance branch is established. Parallel decision is executed under the condition that the first obstacle avoidance branch is the main branch and the second obstacle avoidance branch is the auxiliary branch, and the first obstacle avoidance strategy is determined.

[0055] The static obstacle target refers to the fixed obstacles in the environment, such as walls, columns, obstacle piles, etc. These obstacles are relatively stable and do not change rapidly. For static obstacle targets, the first obstacle avoidance branch (based on hierarchical topological map) is used for decision-making. The first obstacle avoidance branch mainly relies on geometric and spatial segmentation information, which can effectively identify static obstacles and plan an obstacle avoidance path. The first weight is set to a high value, indicating that the first obstacle avoidance branch plays a leading role in the obstacle avoidance process of static obstacles. Since the position and shape of static obstacles are relatively fixed, the topological map can provide relatively stable path planning information. The second obstacle avoidance branch (based on Riemannian manifold graph) works as an auxiliary branch when dealing with static obstacles. Although the Riemannian manifold graph mainly deals with the curvature changes of dynamic obstacles, it can also provide some additional information for static obstacle avoidance. The second weight is low, meaning that the second obstacle avoidance branch contributes less to static obstacle avoidance, but it can provide curvature information to the first obstacle avoidance branch to optimize decision-making.

[0056] The one-way interaction from the second obstacle avoidance branch to the first obstacle avoidance branch means that information is only allowed to pass from the second obstacle avoidance branch to the first obstacle avoidance branch, but not in the reverse direction. The purpose of this is to ensure that the first obstacle avoidance branch can be the dominant decision maker, while the second obstacle avoidance branch provides additional auxiliary information. Parallel decision-making means that both obstacle avoidance branches are running simultaneously. In this way, the first obstacle avoidance branch can make a dominant decision based on the topological map, while the curvature information of the second obstacle avoidance branch serves as an auxiliary decision.

[0057] Further, if the first obstacle target is a dynamic obstacle target, the second weight is identified for the first obstacle avoidance branch, the first weight is identified for the second obstacle avoidance branch, and one-way interaction from the first obstacle avoidance branch to the second obstacle avoidance branch is established. Parallel decision-making with the second obstacle avoidance branch as the main branch and the first obstacle avoidance branch as the auxiliary branch is performed to determine the first obstacle avoidance strategy.

[0058] The motion trajectory of the dynamic obstacle is not fixed, which can affect path selection and obstacle avoidance strategy. For the dynamic obstacle, the weight of the first obstacle avoidance branch is reduced, and the weight of the second obstacle avoidance branch is increased. Specifically, the weight of the first obstacle avoidance branch (topological map) is set to a lower second weight, because the real-time reaction capability of the topological map to the dynamic obstacle is weak. The weight of the second obstacle avoidance branch (Riemannian manifold map) is set to a higher first weight, because the Riemannian manifold map can capture the motion change (such as the change of curvature) of the obstacle in real time and make adjustment. In this way, the second obstacle avoidance branch serves as the main branch, responsible for adjusting the obstacle avoidance path in real time according to the motion trajectory of the dynamic obstacle. The Riemannian manifold map can capture the motion and local curvature change of the obstacle, so as to more effectively deal with the dynamic obstacle. The first obstacle avoidance branch serves as the auxiliary branch, which continues to provide information of static obstacles and overall path planning, but is no longer the main decision maker. The one-way interaction still exists, but in this scenario, the information flow is from the first obstacle avoidance branch to the second obstacle avoidance branch, that is, the information of the topological map assists the Riemannian manifold map to make more accurate dynamic obstacle avoidance path planning. The final strategy is still parallel, but the main branch turns to the manifold map to deal with the dynamic obstacle, and the auxiliary branch is used to supplement the static environment information, so as to ensure that the device can safely avoid obstacles in a complex dynamic environment.

[0059] In summary, the dynamic obstacle avoidance method for the mobile storage and charging device provided by the embodiments of the present application has the following technical effects:

[0060] By explicitly setting the regional installed capacity of the target area, the operation range and device demand can be effectively evaluated, and according to factors such as the actual size of the region and the obstacle density, a suitable path planning and obstacle avoidance strategy can be selected for the mobile storage and charging device. By constructing a dynamically updated hierarchical topological map, accurate path planning can be performed according to different levels of information. Each level of topological map reflects different levels of environmental information, thereby reducing the calculation amount and improving the efficiency and accuracy of path planning. By constructing a dynamically updated Riemannian manifold map, the motion of obstacles is equivalent to the Riemann curvature tensor, so that the system can handle the changes caused by dynamic obstacles and adjust the obstacle avoidance path in real time to ensure that the device can efficiently avoid obstacles. By triggering the obstacle avoidance decision maker, obstacles can be identified in real time during the actual motion of the mobile storage and charging device, and the obstacle avoidance strategy can be adjusted according to different types of obstacles. The dynamic weight setting ensures that in the case of static obstacles and dynamic obstacles, the main branch and the auxiliary branch can be flexibly selected according to the actual situation. This dynamic adjustment avoids overly conservative or overly aggressive obstacle avoidance strategies, thereby improving the efficiency and safety of obstacle avoidance. After the first obstacle avoidance strategy is determined, the walking mechanism of the mobile storage and charging device is adjusted in real time to ensure that the device can travel along the optimal path while avoiding collision with obstacles. Automatic control greatly reduces manual intervention and improves work efficiency and safety.

[0061] Embodiment two, based on the same inventive concept as the dynamic obstacle avoidance method for mobile storage and charging equipment in the preceding embodiment, as Figure 2 As shown in the embodiment of the present application, a dynamic obstacle avoidance system for mobile storage and charging equipment is provided, which comprises:

[0062] A regional installation scale determination module 10 is configured to determine a regional installation scale for a target region of a mobile storage and charging equipment operation; a Riemannian manifold graph construction module 20 is configured to construct a dynamically updated hierarchical topological map and a Riemannian manifold graph according to the regional installation scale, wherein the information mode of each layer of the topological map is different, and the Riemannian manifold graph is equivalent to obstacle movement as a Riemann curvature tensor; an obstacle avoidance strategy determination module 30 is configured to trigger an obstacle avoidance decision maker for a first obstacle target with the movement of the mobile storage and charging equipment, and to determine a first obstacle avoidance strategy according to the branch master-slave deployment and obstacle avoidance decision of the obstacle avoidance decision maker under dynamic weight for static obstacle targets or dynamic obstacle targets, wherein the obstacle avoidance decision maker comprises a first obstacle avoidance branch constructed according to the hierarchical topological map and a second obstacle avoidance branch constructed based on the Riemannian manifold graph; and an automatic obstacle avoidance control module 40 is configured to automatically control the walking mechanism of the mobile storage and charging equipment according to the first obstacle avoidance strategy.

[0063] Further, the Riemannian manifold graph construction module 20 is configured to perform the following operation steps:

[0064] For the target region, a three-dimensional coordinate space is constructed; in the three-dimensional coordinate space, a bottom layer topological map is determined based on grid and point cloud geometric space segmentation; in the three-dimensional coordinate space, a middle layer topological map is determined based on semantic segmentation network and obstacle type annotation; in the three-dimensional coordinate space, a high layer topological map is determined with passable areas as nodes and semantic risk as edge weights; and the bottom layer topological map, the middle layer topological map and the high layer topological map are mapped as the hierarchical topological map according to the space phase.

[0065] Further, the Riemannian manifold graph construction module 20 is configured to perform the following operation steps:

[0066] A preset movement task of the mobile storage and charging equipment is obtained, and a first obstacle distribution of the task trajectory is determined, wherein the first obstacle distribution comprises dynamic obstacle targets and static obstacle targets; for the first obstacle distribution, obstacle movement is equivalent to a Riemann curvature tensor, and a Riemannian manifold graph is constructed.

[0067] Further, a dynamic curvature relationship is determined, and the construction of the Riemannian manifold graph is performed according to the dynamic curvature relationship, wherein the dynamic curvature relationship is that when the obstacle moves, the curvature tensor component of the local manifold around the obstacle changes over time; when the obstacle approaches, the local curvature increases; when the obstacle moves away, the local curvature decreases; and when the curvature exceeds the limit, the obstacle avoidance is triggered.

[0068] Further, the Riemannian manifold graph construction module 20 is configured to perform the following steps:

[0069] With the movement of the mobile charging device, the target obstacle is determined by performing a sensing detection scan through the detection assembly deployed at the front end of the device; it is determined whether the target obstacle belongs to the first obstacle distribution; if not, synchronization is performed in the hierarchical topological map and the Riemannian manifold graph.

[0070] Further, the obstacle avoidance strategy determination module 30 is configured to perform the following steps:

[0071] According to the hierarchical topological map, a first obstacle avoidance branch is constructed; according to the Riemannian manifold graph, a second obstacle avoidance branch is constructed; the first obstacle avoidance branch and the second obstacle avoidance branch are integrated as an obstacle avoidance decision maker and are embedded in the intelligent central control of the mobile charging device.

[0072] Further, the obstacle avoidance strategy determination module 30 is configured to perform the following steps:

[0073] A first obstacle target is determined, wherein the first obstacle target is any one of the first obstacle distribution or the target obstacle; a first weight and a second weight are set, and according to the type of obstacle target, the first obstacle avoidance branch and the second obstacle avoidance branch are dynamically weighted and deployed and obstacle avoidance decisions are made to determine a first obstacle avoidance strategy, wherein the first weight > the second weight.

[0074] Further, if the first obstacle target is a static obstacle target, the first obstacle avoidance branch is identified with the first weight, the second obstacle avoidance branch is identified with the second weight, and a one-way interaction from the second obstacle avoidance branch to the first obstacle avoidance branch is established, and parallel decisions are made with the first obstacle avoidance branch as the main branch and the second obstacle avoidance branch as the auxiliary branch to determine the first obstacle avoidance strategy.

[0075] Further, if the first obstacle target is a dynamic obstacle target, the first obstacle avoidance branch is identified with the second weight, the second obstacle avoidance branch is identified with the first weight, and a one-way interaction from the first obstacle avoidance branch to the second obstacle avoidance branch is established, and parallel decisions are made with the second obstacle avoidance branch as the main branch and the first obstacle avoidance branch as the auxiliary branch to determine the first obstacle avoidance strategy.

[0076] Through the foregoing detailed description of the dynamic obstacle avoidance method for the mobile charging device, those skilled in the art can clearly understand the dynamic obstacle avoidance system for the mobile charging device in the present embodiment. Since it corresponds to the method disclosed in the embodiment, it is described relatively simply, and the relevant part is referred to the method part description.

[0077] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Modifications of these embodiments will occur to persons of skill in the art, and that the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Therefore, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A dynamic obstacle avoidance method for mobile storage and charging equipment, characterized in that, The method includes: Determine the regional installed capacity for the target area of ​​mobile storage and charging equipment operations; Based on the installed capacity of the region, a dynamically updated hierarchical topology map and a Riemann manifold are constructed. The information patterns of each layer of the topology map are different, and the Riemann manifold converts obstacle motion into a Riemann curvature tensor. As the mobile storage and charging device moves, an obstacle avoidance decision-maker is triggered for the first obstacle target. The obstacle avoidance decision-maker performs branch master-slave deployment and obstacle avoidance decision under dynamic weights based on the static or dynamic obstacle target to determine the first obstacle avoidance strategy. The obstacle avoidance decision-maker includes a first obstacle avoidance branch constructed based on a hierarchical topology map and a second obstacle avoidance branch constructed based on a Riemann manifold graph. According to the first obstacle avoidance strategy, the walking mechanism of the mobile storage and charging equipment is automatically controlled to avoid obstacles. Constructing a dynamically updated Riemannian manifold graph includes: Obtain a preset mobile task for the mobile storage and charging device, and determine a first obstacle distribution for the task trajectory, wherein the first obstacle distribution includes dynamic obstacle targets and static obstacle targets; For the first obstacle distribution, the obstacle motion is equivalent to the Riemann curvature tensor, and a Riemann manifold is constructed; After constructing a dynamically updated hierarchical topology map and Riemannian manifold, the following steps are included: As the mobile storage and charging equipment moves, the detection components deployed at the front end of the equipment perform sensing and scanning to identify target obstacles; Determine whether the target obstacle belongs to the first obstacle distribution; If it does not belong to the topology map, it is synchronized with the Riemannian manifold diagram. The obstacle avoidance decision-maker is dynamically weighted for branch master-slave deployment and obstacle avoidance decision-making to determine the first obstacle avoidance strategy, including: Identify a first obstacle target, wherein the first obstacle target is either the first obstacle distribution or any one of the target obstacles; A first weight and a second weight are set. Based on the type of obstacle target, dynamic weight deployment and obstacle avoidance decision are made for the first obstacle avoidance branch and the second obstacle avoidance branch to determine the first obstacle avoidance strategy, wherein the first weight > the second weight; If the first obstacle target is a static obstacle target, the first obstacle avoidance branch is assigned a first weight label, the second obstacle avoidance branch is assigned a second weight label, and a one-way interaction is established between the second obstacle avoidance branch and the first obstacle avoidance branch. Parallel decision-making is performed with the first obstacle avoidance branch as the primary path and the second obstacle avoidance branch as the secondary path to determine the first obstacle avoidance strategy. If the first obstacle target is a dynamic obstacle target, the first obstacle avoidance branch is assigned a second weight label, the second obstacle avoidance branch is assigned a first weight label, and a one-way interaction is established between the first obstacle avoidance branch and the second obstacle avoidance branch. Parallel decision-making is performed with the second obstacle avoidance branch as the primary path and the first obstacle avoidance branch as the secondary path to determine the first obstacle avoidance strategy.

2. The dynamic obstacle avoidance method for mobile storage and charging equipment as described in claim 1, characterized in that, Constructing a dynamically updated hierarchical topology map includes: For the target region, a three-dimensional coordinate space is constructed; In the three-dimensional coordinate space, the underlying topological map is determined by geometric spatial segmentation based on grid and point cloud. In the three-dimensional coordinate space, a mid-level topology map is determined based on the semantic segmentation network and obstacle type annotations; In the three-dimensional coordinate space, a high-level topology map is determined using traversable areas as nodes and semantic risks as edge weights. The bottom-level topology map, the middle-level topology map, and the high-level topology map are used as the hierarchical topology map based on spatial phase mapping.

3. The dynamic obstacle avoidance method for mobile storage and charging equipment as described in claim 1, characterized in that, Determine the dynamic curvature relationship, and construct the Riemannian manifold based on the dynamic curvature relationship, wherein the dynamic curvature relationship is the change of the curvature tensor components of the local manifold around the obstacle with time when the obstacle moves; When an obstacle approaches, the local curvature increases; when an obstacle moves away, the local curvature decreases. When the curvature exceeds the limit, obstacle avoidance is triggered.

4. The dynamic obstacle avoidance method for mobile storage and charging equipment as described in claim 1, characterized in that, Before triggering the obstacle avoidance decision-maker for the first obstacle target, the construction of the obstacle avoidance decision-maker includes: Based on the hierarchical topology map, construct the first obstacle avoidance branch; Based on the Riemannian manifold diagram, construct the second obstacle avoidance branch; The first obstacle avoidance branch and the second obstacle avoidance branch are integrated to serve as an obstacle avoidance decision-maker, and are embedded in the intelligent central control of the mobile storage and charging equipment.

5. A dynamic obstacle avoidance system for mobile storage and charging equipment, characterized in that, The system is used to implement the dynamic obstacle avoidance method for mobile storage and charging equipment according to any one of claims 1-4, the system comprising: The regional installed capacity determination module is used to determine the regional installed capacity for the target area where mobile storage and charging equipment is operated. The Riemann manifold construction module is used to construct a dynamically updated hierarchical topology map and a Riemann manifold map based on the installed capacity of the region. The information patterns of each layer of the topology map are different, and the Riemann manifold map equates obstacle motion to the Riemann curvature tensor. The obstacle avoidance strategy determination module is used to trigger the obstacle avoidance decision-maker for the first obstacle target as the mobile storage and charging device moves. The obstacle avoidance decision-maker performs branch master-slave deployment and obstacle avoidance decision under dynamic weights according to the static or dynamic obstacle target to determine the first obstacle avoidance strategy. The obstacle avoidance decision-maker includes a first obstacle avoidance branch constructed based on a hierarchical topology map and a second obstacle avoidance branch constructed based on a Riemann manifold graph. The automatic obstacle avoidance control module is used to automatically control the walking mechanism of the mobile storage and charging device according to the first obstacle avoidance strategy.

Images