CN116310743A

CN116310743A - Method, device, mobile device and storage medium for determining expansion strategy

Info

Publication number: CN116310743A
Application number: CN202310181519.3A
Authority: CN
Inventors: 请求不公布姓名
Original assignee: Beijing Idriverplus Technologies Co Ltd
Current assignee: Beijing Idriverplus Technologies Co Ltd
Priority date: 2023-02-20
Filing date: 2023-02-20
Publication date: 2023-06-23

Abstract

The application relates to a method, an apparatus, a mobile device and a storage medium for determining an expansion policy. The method for determining the expansion strategy comprises the following steps: obtaining a plurality of elements, wherein each element has corresponding cost information and semantic information, and a corresponding relation exists between the semantic information and the expansion strategy; mapping a plurality of elements to an expansion layer respectively to obtain cost information and semantic information of each unit in the expansion layer; and determining the expansion strategy of each grid in the expansion layer based on the corresponding relation and the cost information and semantic information of each grid in the expansion layer. The scheme that this application provided can promote the precision of inflation processing result.

Description

Method, device, mobile device and storage medium for determining expansion strategy

Technical Field

The present application relates to the field of computer-assisted and/or autopilot technology, and more particularly, to a method, apparatus, mobile device and storage medium for determining an inflation strategy.

Background

With the continuous development of electronic map and vehicle navigation technologies, vehicles with computer-aided and/or automatic driving technologies (hereinafter referred to as vehicles) have entered people's daily lives. The vehicle needs to plan a path based at least on the map information to reach the destination based on the path.

A plurality of sensing modules may be provided on the vehicle to sense a plurality of elements to determine obstructions, etc. The elements are projected to the layer of the map, so that the path planning based on the real-time environment can be realized. The related art performs expansion processing on an area where an obstacle is located in a map, etc., to increase the security of the mobile device. However, the accuracy of the expansion processing result is to be improved.

Disclosure of Invention

In order to solve or partially solve the problems in the related art, the application provides a method, a device, a mobile device and a storage medium for determining an expansion strategy, which can effectively improve the accuracy of an expansion processing result.

In one aspect, the present application provides a method of determining an expansion strategy, comprising: obtaining a plurality of elements, wherein each element has corresponding cost information and semantic information, and a corresponding relation exists between the semantic information and the expansion strategy; mapping various elements to a main layer respectively to obtain cost information and semantic information of each unit in the main layer; and determining the expansion strategy of each grid in the main layer based on the corresponding relation and the cost information and semantic information of each grid in the main layer.

Another aspect of the present application provides an apparatus for determining an expansion strategy, comprising: the element obtaining module is used for obtaining a plurality of elements, wherein each element has corresponding cost information and semantic information, and a corresponding relation exists between the semantic information and the expansion strategy; the element mapping module is used for mapping various elements to the main layer respectively to obtain cost information and semantic information of each unit in the main layer; and the expansion strategy determining module is used for determining the expansion strategy of each grid in the main layer based on the corresponding relation and the cost information and the semantic information of each grid in the main layer.

Another aspect of the present application provides a mobile device, comprising: a mobile device body; the sensing device is arranged on the mobile device main body and is used for obtaining motion related information of the mobile device main body; a controller, including at least one processor, coupled to the sensing device for processing motion related information; a memory having executable code stored thereon which, when executed by a processor, causes the processor to perform a method as described above.

Another aspect of the present application provides a computer readable storage medium having executable code stored thereon, which when executed by a processor of an electronic device, causes the processor to perform a method as above.

Another aspect of the present application provides a computer program product comprising executable code which, when executed, implements a method as above.

Another aspect of the present application provides a controller comprising: the system comprises at least one processor and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the method as described above.

The technical scheme that this application provided can include following beneficial effect:

according to the technical scheme, the obtained multiple elements comprise cost information and semantic information, and a corresponding relation exists between the semantic information and the expansion strategy. Therefore, the cost information and the semantic information of each unit of the main layer can be obtained through the modes of mapping, fusion and the like, and the corresponding expansion strategy is further determined. According to the embodiment of the application, different expansion strategies are provided for various elements respectively, so that the aim of classified expansion is fulfilled, and the accuracy of an expansion result is improved.

According to the technical scheme, cost information and semantic information are represented by the numerical value with the preset byte length, so that the consumption of computing resources is reduced on the basis that the accuracy of an expansion result can be improved, and the response speed is improved.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.

Drawings

The foregoing and other objects, features and advantages of the application will be apparent from the following more particular descriptions of exemplary embodiments of the application as illustrated in the accompanying drawings wherein like reference numbers generally represent like parts throughout the exemplary embodiments of the application.

FIG. 1 is an exemplary system architecture of a method, apparatus, mobile device, and storage medium to which a determine inflation policy may be applied, as illustrated in one embodiment of the present application;

FIG. 2 is a schematic diagram of a vehicle, onboard sensor, as shown in an embodiment of the present application;

FIG. 3 is a flow chart illustrating a method of determining an expansion strategy according to an embodiment of the present application;

FIG. 4 is a schematic diagram showing a correspondence relationship between cost value and mobile device size according to an embodiment of the present application;

FIG. 5 is a schematic illustration of the expansion results shown in an embodiment of the present application;

FIG. 6 is a schematic representation of cost information and semantic information according to one embodiment of the present application;

FIG. 7 is a schematic diagram of cost information and semantic information as shown in an embodiment of the present application;

FIG. 8 is another schematic diagram of cost information and semantic information shown in an embodiment of the present application;

FIG. 9 is a schematic diagram of projecting an intumescent layer or the like onto a primary layer, as shown in an embodiment of the present application;

FIG. 10 is a block diagram of a determination of an inflation strategy according to one embodiment of the present application; and

fig. 11 is a block diagram of a mobile device according to an embodiment of the present application.

Detailed Description

Embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While embodiments of the present application are shown in the drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

The terminology used in the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used in this application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any or all possible combinations of one or more of the associated listed items.

It should be understood that although the terms "first," "second," "third," etc. may be used herein to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first message may also be referred to as a second message, and similarly, a second message may also be referred to as a first message, without departing from the scope of the present application. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. In the description of the present application, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.

The term "mobile device" as used herein includes, but is not limited to, vehicles of the L0-L5 automated driving technical class defined by the International Association of automaton engineers (Society of Automotive Engineers International, SAE International for short) or the national Standard for automotive Automation Classification in China.

In some embodiments, the mobile device may be a vehicle device or a robotic device having various functions:

(1) Manned functions such as home cars, buses, etc.;

(2) Cargo functions such as common trucks, van type trucks, swing trailers, closed trucks, tank trucks, flatbed trucks, container trucks, dump trucks, special structure trucks, and the like;

(3) Tool functions such as logistics distribution vehicles, automatic guided vehicles AGVs, patrol vehicles, cranes, excavators, bulldozers, shovels, road rollers, loaders, off-road engineering vehicles, armored engineering vehicles, sewage treatment vehicles, sanitation vehicles, dust collection vehicles, floor cleaning vehicles, watering vehicles, floor sweeping robots, meal delivery robots, shopping guide robots, mowers, golf carts, and the like;

(4) Entertainment functions such as recreational vehicles, casino autopilots, balance cars, etc.;

(5) Special rescue functions such as fire trucks, ambulances, electric power emergency vehicles, engineering emergency vehicles and the like.

For convenience of description, embodiments of the present invention will be described with reference to a mobile device as a vehicle.

With the rapid development of advanced driving assistance systems (Advanced Driving Assistance System, abbreviated as ADAS) and unmanned technologies, various sensors are loaded on automobiles to sense the surroundings of the automobiles. For example, the recognition of the vehicle to the external environment is completed by a sensing module, and the sensing module outputs the recognized surrounding vehicles, pedestrians, trees, buildings and the like through the algorithm processing by accessing the data of various sensors. The data sources of the sensing module are various, including cameras, laser radars, millimeter wave radars, satellite navigation and the like.

For example, during daily driving, a camera acts as a sensor for capturing two-dimensional images within a specific range of viewing angles. Automatic driving is then achieved/assisted based on image recognition techniques. Among them, object detection is an important content of the automatic driving technology, and it is necessary to accurately detect all objects on a road surface and ensure driving safety.

For example, a laser radar (Laser Detecting and Ranging, abbreviated as Lidar) determines a distance by transmitting and receiving a laser beam, measuring a time difference and a phase difference of a laser signal, measuring an angle by horizontal rotation scanning, establishing a two-dimensional polar coordinate system according to the two parameters, and acquiring height information in three dimensions by acquiring signals of different pitching angles. High frequency lasers can acquire a large amount (e.g., 150 tens of thousands) of positional point information (referred to as point cloud) within one second, and perform three-dimensional modeling based on the information. The laser radar has the advantages of high resolution, strong active interference resistance, good low-altitude detection performance, light weight, flexibility and the like.

The device (mobile apparatus with storage computing medium) (mobile robot or vehicle) recognizes the environment through various types of sensors while autonomously navigating in the indoor and outdoor environments. The prior map information, the area information and the sensor perception data are represented by different layers of the cost map, projected onto a main layer and expanded based on the equipment model.

For example, the related art may capture an image by a camera, obtain point cloud data by radar, identify an obstacle by an image identification technique and/or a point cloud identification technique, and the like. Then, the point cloud data, the image data, the recognition result, and the like are respectively used as a map layer, projected onto the main layer, and then subjected to expansion processing.

The low cost of the related art reflects the environmental change technical scheme, and one possible implementation is a cost map. Map and perception data in a certain range are described by a two-dimensional array through mapping from the obstacle points to the map grids, and expansion is carried out on the obstacle points based on the vehicle model so as to support safety verification during path searching.

However, in the cost map scheme of the related art, the cost values of the map layers are sequentially updated to the main cost map layer by the plurality of map layers to be overlapped, the expansion layer expands based on the cost of the obstacle point of the main layer, and the expansion process uses a uniform expansion radius and attenuation coefficient. The applicant finds that the same expansion strategy is adopted for expansion operation of different objects, so that the confidence difference between different map layers cannot be reflected, and the precision of expansion results is low.

According to the embodiment of the application, the cost map for classified expansion can be realized, various element information (such as point cloud data, image data and the like) is recorded and transmitted, different expansion strategies are executed on the obstacle points on the main layer, the confidence difference of different elements is embodied, and the equipment operation is better guided.

A method, apparatus and electronic device for storing video according to embodiments of the present application will be described in detail below with reference to fig. 1 to 11.

FIG. 1 is an exemplary system architecture of a method, apparatus, mobile device, and storage medium that may apply a determined expansion policy, as illustrated in an embodiment of the present application. It should be noted that fig. 1 is only an example of a system architecture to which the embodiments of the present application may be applied to help those skilled in the art understand the technical content of the present application, and does not mean that the embodiments of the present application may not be used in other devices, systems, environments, or scenarios.

Referring to fig. 1, a system architecture 100 according to this embodiment may include

mobile platforms

101, 102, 103, a network 104, and a cloud 105. The network 104 is the medium used to provide communication links between the

mobile platforms

101, 102, 103 and the cloud 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others. The

mobile platforms

101 and 102 may be equipped with sensors, such as cameras, lidar, millimeter wave radar, etc., to sense information about the surrounding environment of the

mobile platforms

101 and 102, such as obstacle information, etc.

A user may interact with other mobile platforms and cloud 105 over network 104 using

mobile platforms

101, 102, 103 to receive or transmit information, etc. Such as sending model training requests, downloading model parameters requests, and receiving trained model parameters. The trained models may identify objects in image data or point cloud data, such as obstacles, traffic indication signs, and the like. The

mobile platforms

101, 102, 103 may be installed with various communication client applications, such as, for example, driving assistance applications, autopilot applications, vehicle applications, web browser applications, database class applications, search class applications, instant messaging tools, mailbox clients, social platform software, and the like.

Mobile platforms

101, 102, 103 include, but are not limited to, automotive, robotic, aircraft, tablet, laptop portable, and like electronic devices.

The cloud 105 may receive a model training request, a model parameter downloading request, etc., adjust model parameters to perform model training, issue model topology, issue trained model parameters, issue updated applications, etc., and may also send weather information, real-time traffic information, etc., to the

mobile platforms

101, 102, 103. For example, the cloud 105 may be a background management server, a server cluster, a car networking, or the like.

It should be noted that the number of servers in the mobile platform, network and cloud are merely illustrative. There may be any number of mobile platforms, networks, and clouds, as desired for implementation.

Fig. 2 is a schematic structural view of a vehicle-mounted sensor according to an embodiment of the present application.

Referring to fig. 2, a vehicle 200 may include one or more of the following components: a body main body 110, a camera 120, and a laser radar 140. The photographing device 120 is disposed on the body 110. In addition, in order to improve the driving assistance or the automatic driving performance, an inertial measurement unit (Inertial Measurement Unit, abbreviated as IMU) may be further provided on the vehicle 200, and the IMU is provided in the body main body 110.

For example, the photographing device 120 includes a camera, and the optical signal collected by the camera is imaged on the photosensor and transmitted to the image processor. In order to ensure that a clear picture can be taken during driving, camera pixels include, but are not limited to: 30 ten thousand, 100 ten thousand, 200 ten thousand, 500 ten thousand or 800 ten thousand, etc. The image quality resolution of the presentation includes 360P, 720P, 1080P, 2K, or even 4K. Aiming at different light environments, clear shooting under the conditions of low light level, weak light and even backlight can be realized through modes such as mode adjustment or light filling. Meanwhile, the high frame rate function also ensures that pictures shot under the high-speed driving condition are still reliable. The image information may be used as an element for constructing a cost map.

For example, the lidar 140 may be mounted at a roof position, a front end of the vehicle, or a position above the front window of the vehicle, and the like, without limitation. The lidar 140 may include a pulse lidar and a continuous wave lidar, and ultraviolet, visible or near infrared light may be used as a light source. The point cloud data may be used as an element for constructing a cost map.

FIG. 3 is a flow chart illustrating a method of determining an expansion strategy according to an embodiment of the present application.

Referring to fig. 3, the method of determining an expansion policy may include operations S310 to S330.

In operation S310, a plurality of elements are obtained, each element having corresponding cost information and semantic information, and a correspondence exists between the semantic information and the expansion policy.

In this embodiment, an element may include information from one or more sources. Sources include, but are not limited to, cameras, radars, interactive interfaces, and the like. For example, the image information may be an element. For example, the point cloud data may be an element. For example, a static map may be an element. For example, the manually set region may be an element. For example, the image information and the point cloud data may be taken together as one element. In a specific embodiment, the elements include: at least one of a priori map, area information, sensor class, semantic output, a plurality of elements may be used in combination.

The cost information is a probability that a collision of the mobile device at a certain location can be characterized. For example, the cost information may be represented by a cost value, a character string, or the like. Specifically, the cost information is characterized by a numerical value, and the numerical value of the specific value range corresponds to the specific collision risk probability range. For example, if cost information is represented by 4-bit binary data, the maximum value 15 may indicate that a collision must occur, 0 may indicate that no collision must occur, and the cost value therebetween indicates that there is a certain collision probability. With the cost information represented by 8-bit binary data, a maximum value 255 may indicate that a collision must occur, a 0 may indicate that no collision must occur, and a cost value therebetween indicates that there is a certain collision probability.

The semantic information may characterize the artifacts or confidence of the corresponding location of the current grid. For example, if the semantic information characterizes that the element is point cloud data, the confidence of the element is high. For example, the semantic information characterizes that the element is image information, the confidence of the image information is lower than the confidence of the point cloud data. For example, the semantic information characterizes that the element is a small animal, and the element has a larger expansion radius than a common obstacle. For example, the semantic information characterizes that the element is a child, and the expansion radius of the element may be set to be larger than the expansion radius of a small animal.

The inflation policy may be related to cost information. For example, the expansion strategy determines the expansion radius and the change rule of the cost value, such as linear transformation or nonlinear change. In particular, the parameters of the expansion strategy may include at least one of an expansion radius and an attenuation coefficient.

In some embodiments, the expansion can be performed by combining information such as semantics, prior probability and the like on the point cloud layer surface without using a cost map. The implementation increases the point cloud output, and simultaneously, both the transmitting end and the receiving end increase the data volume, thereby increasing the calculation burden.

In some embodiments, the above method may further comprise the following operations: and constructing map layers corresponding to each element, wherein each map layer comprises a plurality of grids, and corresponding cost information and semantic information exist in each grid. Thus, the cost map is convenient to be used for fusing and expanding various elements.

The cost map of the single layer is divided into a plurality of layers, so that one map layer can be updated by only using the same data. For example, static map layers are map layers generated using static map data. The sensing layer is a map layer updated by the sensor data. A map layer may also be customized for some specific purpose so that the generated path avoids some areas.

In operation S320, a plurality of elements are mapped to the expansion layer, respectively, to obtain cost information and semantic information of each unit in the expansion layer.

Robot navigation (in particular path planning modules) is map-dependent. A map in a robot operating system (Robot Operating System, ROS for short) may be regarded as a picture with metadata of width, height, resolution, etc., in which a gray value is used to represent the probability of the presence of an obstacle.

The map constructed based on the instant positioning and map construction (Simultaneous Localization and Mapping, SLAM for short) technology is that the robot starts to move from an unknown position in an unknown environment, self-positioning is carried out according to the position and the map in the moving process, and meanwhile, an incremental map is constructed on the basis of self-positioning, so that autonomous positioning and navigation of the robot are realized.

The map constructed by SLAM is a static map, and in the navigation process, the obstacle information is variable, and possibly the obstacle is removed, or a new obstacle is added, so that the obstacle information is required to be acquired from time to time in the navigation process. For example, when approaching the edge of an obstacle, the robot, although here in a free area, may collide with the obstacle after entering the area due to other factors (e.g., inertia, irregularities, robot turns, etc.). In addition, the user may need to set some specific no-pass regions.

Some auxiliary information map, such as obstacle data acquired from time to time, can be added on the basis of the static map, and the data of expansion areas and the like added on the basis of the static map.

The expansion area is exemplarily described by taking an expansion Layer (expansion Layer) in the cost map as an example. The inflated layer in this embodiment includes occupancy grids, cost information, and semantic information for characterizing buffers around objects, where objects include moving obstacles and fixed obstacles. The intumescent layer may provide a buffer around each object. The expansion layer performs the expansion process by inserting a buffer (buffer grid) around the obstacle, the size of which (e.g. buffer radius, which can be expressed by the number of buffer grids) can be related to the semantic information.

In particular, the expansion layer may be expanded (outwardly expanded) over the static map layer and the obstacle map layer to avoid the robot from hitting an obstacle. The expansion radius (expansion radius) is the size of the diameter of the robot chassis that the expansion layer will expand the obstacle cost up to that radius. For example, the location of a positive collision may be marked with a fatal cost, while there is a non-fatal cost. For example, the size of the buffer zone of the moving obstacle and the fixed obstacle are different. The introduction of the expansion map helps to reduce the probability of collision between the robot and the obstacle.

It should be noted that the cost map includes a global cost map and a local cost map. The former is used for global path planning and the latter is used for local path planning. Take the local cost map as an example for illustration. The local cost map may include a map layer other than the expansion layer as shown below.

A Static Map Layer (Static Map Layer) is a substantially constant Map Layer, such as a Static Map built based on SLAM. The static map layer is an occupancy grid that is predetermined a priori due to static objects on or along the road. The static map layer includes cost values and semantic information (e.g., fixed obstacles such as walls, semi-static obstacles such as tables, sliding doors, etc.) of various static and/or semi-static objects (e.g., roadside infrastructure, buildings, etc.), and is used for global planning. Since the static map is the bottom layer of the global hierarchical cost map, the values in the static map can be copied directly into the main cost map (hereinafter referred to as the main layer). If the robot is running SLAM while navigating using the generated map, the layered cost map method allows static map layer updates without losing information in other layers. In a single layer cost map, the entire cost map will be overwritten. Other layers of the hierarchical cost map maintain costs due to dynamic objects as well as security and personal privacy requirements of those objects.

And the obstacle map layer (Obstacle Map Layer) is used for dynamically recording obstacle information perceived by the sensor. The perceived obstacle layers include occupancy grids and cost values representing road and roadside obstacles (e.g., dynamic or static). The perceived obstacle layer determines perceived objects, which are obstacles to be considered during operation, such as during driving. The perceived obstacle layer collects sensor data from a perceived device (e.g., lidar, depth camera, etc.), and stores the collected sensor data in association with a grid of the map. For example, the space between the sensor and the sensor reading is marked as "free-passing" and the location of the sensor reading is marked as "occupied". The method for combining the values of the perceived obstacle layer with the values already in the map layer may vary depending on the sensor data and/or the required level of trust of the sensor type. In some implementations, static map data may be overlaid by collected sensor data, which may be beneficial for scenes where static maps may be inaccurate. In other implementations, the barrier layer may be configured to add deadly barriers to the primary layer only.

Sonar layers, including occupancy grids and cost values representing objects and/or perceived environments detected using sonar sensors and associated circuitry. This layer may be used to detect transparent objects (e.g., glass structures) that may not be detected using other sensors (e.g., lidar, etc.).

Other Layers (Other Layers), other map Layers implemented by plug-ins, such as sensor map Layers, custom Layers, semantic map Layers, and the like. The map layers can be freely matched as required.

The custom layer is based on some map data set by the service itself, for example, a virtual wall which is not built before SLAM mapping can be added in the navigation process.

The semantic map layer is a recognition result obtained by analyzing the data of one or more map layers (such as image recognition, point cloud recognition, sound wave recognition and the like), and the recognition result can represent that some areas have barriers or are inconvenient to pass. For example, the image recognition may be a recognition performed locally for determining whether an obstacle is included in the target image. In addition, obstacles such as pedestrians, vehicles, traffic lights, etc. may be subdivided. Image recognition may be performed by pre-trained classification models including, but not limited to: linear regression (Linear Regression), logistic regression (Logistic Regression), decision Trees (Bayes), naive Bayes, K-nearest neighbor (K-Nearest Neighbors), support vector machine (Support Vector Machines), random Decision forest (Random Decision Forests or Bagging), neural Networks (Neural Networks), etc.

And a plurality of map layers are used for respectively representing different elements, so that the source of each data is conveniently determined. The accuracy of identification based on the same kind of data is higher, and the elements are conveniently mapped to the expansion layer respectively. Specifically, mapping the multiple elements to the expansion layer respectively, and obtaining cost information and semantic information of each unit in the expansion layer may include: and mapping the map layers to the expansion layer to obtain cost information and semantic information of each grid in the expansion layer.

For example, for the cost value, the cost value of the corresponding grid in the expansion layer may be obtained by means of accumulation. For example, semantic information with higher confidence or higher security of the inflated policy may be retained for semantic information.

It should be noted that one or more map layers map to an expansion layer where expansion is based on different semantics and preserve costs and semantics. In addition, the expansion layer and other map layers may also be mapped to the main layer for external querying. The expansion layer is used for expansion algorithm, and even can be not configured, so that the cost map is not expanded, but the main layer is laminated based on other map layers.

In operation S330, an expansion policy of each grid in the expansion layer is determined based on the correspondence and the cost information and semantic information of each grid in the expansion layer.

In this embodiment, there is a correspondence between semantic information and expansion policy, and the expansion radius for the deadly obstacle, the expansion radius for the non-deadly obstacle, the expansion radii corresponding to different elements, and the attenuation coefficients for the cost value of different sections within the expansion radius for the non-deadly obstacle may be determined based on the cost information and the semantic information.

Fig. 4 is a schematic diagram showing a correspondence relationship between a cost value and a mobile device size according to an embodiment of the present application. In fig. 4, a number of parameters involved in the expansion strategy are illustrated by way of example in which the cost value is represented by a four-bit binary value.

Referring to fig. 4, the lower half shows the center of the robot, the outline shape of the robot, inscribed circles, circumscribed circles, inscribed regions, and the like. It can be seen that the risk of interference between the sides of the robot and a particular obstacle is different when the robot is not a circular device. If the robot edge does not collide with the obstacle, the protruding sharp corner portion thereof may interfere with the obstacle. Therefore, the areas such as the center, the inscribed circle, the circumscribed circle and the like of the robot are defined, and the different areas have different probabilities of interference with the obstacle or not, namely have different cost values. The higher the cost value, the higher the probability of interference between the robot and the obstacle.

The upper half of fig. 4 shows the correspondence between the cost value and each region. When the center position of the robot and the obstacle position overlap, interference is inevitably generated, and the robot is defined as a fatal obstacle, for example, the cost value is 15 (the range of the four-bit binary value is 0-15). The inscribed circle area of the robot corresponds to the inscribed obstacle space, and the cost value of the inscribed circle area is slightly lower than that of the center of the robot, for example, the cost value is 14.

For the interval region between the inscribed circle and the circumscribed circle of the robot, whether the robot interferes with the obstacle or not depends on the current posture of the robot. The probability of interference is lower in the region closer to the circumscribed circle among the above-mentioned interval regions. The corresponding interval can be set to characterize the probability of interference, such as the cost value set to the value range of [7,13 ]. The cost value can be changed linearly or nonlinearly in the process of increasing from 7 to 13. The correspondence between the cost value and the specific position in the above-mentioned interval region may be fixed or determined according to a preset rule. The relevant information may be set in the inflation strategy to adjust information such as inflation radius, attenuation rate, etc.

It should be noted that the manner in which the cost value is represented by the four-bit binary number is merely an example, and the cost value may be represented by other manners. Such as represented by an eight bit binary value. Accordingly, the grid value of the deadly obstacle is 254, and at this time, the obstacle overlaps with the center of the robot, and a collision is inevitably generated. The grating value of the inscribed obstacle is 253, and the obstacle is positioned in the inscribed circle of the robot at the moment, so that collision is inevitably generated. The grid value of the circumscribed obstacle is [128,252], and the obstacle is positioned in the circumscribed circle of the robot and is positioned at the collision critical point, so that the obstacle is not necessarily collided. The grid value of the non-free space is (0,127), at this time the robot is near an obstacle, belongs to a dangerous alert zone, enters the zone, and may collide in the future, the grid value of the free zone is 0, and the robot can freely pass through the zone.

In some embodiments, after determining the expansion policy, one or more map layers may be expanded based on the expansion policy, resulting in expansion results, and projected onto the main layer. Therefore, the data of the main layer is affected by expansion operation, and expansion results processed by different expansion strategies can be provided for the outside.

Fig. 5 is a schematic diagram of an expansion result shown in an embodiment of the present application.

Referring to fig. 5, a normal expansion region and a high expansion region are shown. Wherein the expansion rate of the high expansion region is greater than that of the normal expansion region. If the normal expansion region expands outwardly by a first number of cells, the high expansion region expands outwardly by a second number of cells, the second number being greater than the first number. Taking an indoor robot as an example, the high expansion area corresponds to the area where the child is located, and the normal expansion area corresponds to the area where the wall is located. The robot collides with children and possibly causes the children to be injured, and the corresponding expansion strategy is a high expansion strategy, so that the robot is far away from the children as much as possible, and the damage to the outside or the damage to the robot possibly caused by the robot is reduced.

In some embodiments, multiple map layers exist for a common cost map. Accordingly, mapping a plurality of map layers to an expanded layer includes: each map layer of the plurality of map layers is mapped to an inflated layer of the common cost map, respectively.

For example, for each target grid in the expansion layer, mapping each map layer of the plurality of map layers to the expansion layer of the common cost map, respectively, may include the following operations.

First, a plurality of pieces of cost information of grids in a plurality of map layers corresponding to the target grid are overlapped to obtain updated cost information of the target grid, and semantic information with high priority in a plurality of pieces of candidate semantic information is reserved.

Then, the cost information of the target grid is updated based on the updated cost information, and the semantic information of the target grid is updated based on the semantic information with high priority.

In a specific embodiment, to support different expansion strategies, when each map layer updates the cost value of the main layer, the semantic information of each map layer is updated at the same time. The semantic information of each map layer includes, but is not limited to, at least one element of a priori map, area information, sensor category or semantic information, etc. so as to adapt to the actual working environment of different devices (such as robots). For example, for an indoor floor-cleaning vehicle, the sensor layout may greatly affect the equipment operation, and different expansion strategies may be implemented according to the sensor class, and the confidence level of the expansion strategies may be distinguished. For outdoor traffic, different areas (such as overpasses, steep slopes up and down) may greatly affect the movement behavior of the equipment, and expansion strategies of different obstacles may be performed according to the area information. After the expansion information of the elements is recorded, the highest priority or confidence is selected for reservation, and when the main layer obstacle points are expanded, a corresponding expansion strategy is selected, including selecting a proper expansion radius and attenuation coefficient.

In some embodiments, the foregoing expanding layer that maps each of the plurality of map layers to the common cost map, respectively, may include the following operations: for each map layer, the cost information and the semantic information corresponding to each grid in the expansion layer are updated based on the cost information and the semantic information of each grid in the current map layer.

Specifically, updating the cost information and the semantic information corresponding to each grid in the expansion layer based on the cost information and the semantic information of each grid in the current map layer may include the operations of: and for each target grid in the expansion layer, superposing the cost information of the grid in the current map layer corresponding to the target grid with the cost information of the target grid, and updating the cost information of the target grid. And/or retaining semantic information with high priority in the first semantic information and the second semantic information, wherein the first semantic information is the semantic information of the target grid, and the second semantic information is the semantic information of the grid in the current map layer corresponding to the target grid.

For ease of understanding, the expansion strategy is illustrated with area information as an example, see fig. 5, where there are high expansion areas and normal expansion areas in the map. The high expansion region represents a dangerous area where obstacles require a further expansion radius and a lower attenuation coefficient. In the cost map updating process, besides the updating cost value, the area information of the map points is recorded, and a larger value is required to be set for representing because the priority of the high expansion area is higher. In order to embody the superposition and screening strategies, semantic information is reserved, such as temporary barriers in a high expansion area, wherein the temporary barrier expansion strategy is 1, the high expansion area expansion strategy is 2, the high expansion area strategy is reserved preferentially when the main layer is updated, and the temporary barrier semantic strategy is ignored.

The following exemplifies the representation of cost information and semantic information.

In some embodiments, the cost information and the semantic information are characterized by a value having a predetermined byte length, and a sum of the byte length of the cost information and the byte length of the semantic information is less than or equal to the predetermined byte length. The predetermined byte length may be set according to an actual precision requirement, such as a precision requirement, the number of element types, the number of security levels, and the like. For example, the predetermined byte length is 4 bits, 8 bits, 12 bits, 16 bits or more, or the like. The cost information and the semantic information may occupy at least part of bits in the value of the predetermined byte length.

In some embodiments, the cost information is characterized using a first portion of bits in 1 bits and the cost information is characterized using a second portion of bits in 1 bits. Wherein the first part of bits and the second part of bits may be different bits. The number of bits occupied by each may be the same or different.

It should be noted that the first portion of bits and the second portion of bits may be isolated from each other or overlap. Taking the example of cost information and semantic information represented by 8-bit binary data. The first five bits represent cost information and the second four bits represent semantic information. The specific value of the fifth bit in 8 bits is based on semantic information. For example, a five-bit binary value maximum number of expressions of 32 can express more cost values than a four-bit binary value maximum number of expressions of 16. Wherein 0 represents free passage, and 1 to 31 represent a certain interference probability. The last inaccurate influence range in the five-bit binary number value is 1, and no larger influence is brought. However, the four-bit semantic information and the three-bit semantic information are respectively 16 kinds and 8 kinds of semantic types which can be expressed, and the difference is large. On the basis of not influencing the cost value accuracy excessively, the variety number of the cost information and the semantic information which can be expressed is obviously increased.

Fig. 6 is a schematic representation of cost information and semantic information according to an embodiment of the present application.

Referring to fig. 6, three representations for cost information and semantic information are shown by way of example. The use of 8-bit binary data to characterize cost information and semantic information is shown in fig. 6 (a). Wherein, the cost information occupies the first four bits, and the semantic information occupies the last four bits. As can be seen in connection with fig. 4, if the cost information is 0000, this indicates that free traffic is possible. If the cost information is 0010 and the semantic information is 0001, the cost value is 2, and the semantic information can represent a low expansion policy. After the information is determined, expansion processing can be performed on the current map layer based on the perceived position of the obstacle, the cost value and the low expansion strategy, and an expansion result is obtained.

Wherein the semantic information may include 16 kinds. When the map layer is projected to the expansion layer and the semantic information is updated, the overlay can be performed according to the confidence level. For example, lidar has a higher confidence than camera, and when outputting a cloud of points at the same location, the lidar semantics override the camera semantics, which are performed in accordance with the lidar's strategy when inflated. When updating cost information, a higher cost value can be reserved, and the semantics of a sensor with the higher cost value can be reserved.

The use of 8-bit binary data to characterize cost information and semantic information is shown in fig. 6 (b). Wherein, the cost information occupies the first five bits, and the semantic information occupies the last three bits. The difference compared to fig. 6 (a) is the number of bits each occupied by the cost information and the semantic information is different. This allows the user to adjust the accuracy of the cost information and the semantic information based on the characteristics of the current scene. For example, the more bits that are occupied, the more kinds that can be expressed, the more cost values can be used to more carefully characterize the risk of collision between the robot and the obstacle. In addition, for some scenes with fewer semantic information types, more bits can be allocated to the cost information by reducing the number of bits used for representing the semantic information, so that the accuracy of the cost information is improved, and vice versa.

In fig. 6 (a) and 6 (b), the bit of the cost information is the front bit, the bit of the semantic information is the back bit, and the bit of the cost information is the back bit, and the bit of the semantic information is the front bit may be adjusted. The cost information and the semantic information occupy all bits in fig. 6 (a) and 6 (b), which are only examples, and only a part of bits may be occupied. The cost information and the semantic information in fig. 6 (a) and 6 (b) each occupy consecutive bits, which is only an example, and may each occupy scattered bits. For example, the first bit of the semantic information, when 1, characterizes the low expansion region, and the second bit, when 1, characterizes the high expansion region.

In summary, the representation modes of the cost information and the semantic information can be selected according to different scenes or behavior constraint schemes.

The use of 12-bit binary data to characterize cost information and semantic information is shown in fig. 6 (c). Wherein, the cost information occupies the first 8 bits, and the semantic information occupies the last four bits. The difference compared to fig. 6 (a) is the number of bits occupied by the cost information. Therefore, a proper representation form can be selected according to the requirement of the time application scene on the precision, and the consumption of calculation resources is reduced as much as possible on the basis of meeting the requirement on the precision.

The process of mapping a plurality of map layers to an expanded layer or a main layer is exemplarily described below.

Fig. 7 is a schematic diagram of cost information and semantic information according to an embodiment of the present application.

Referring to fig. 7, for a certain specific object, cost information and semantic information of a certain map layer are respectively expressed as: 1111 and 0001. Wherein, the cost value 1111 has a value of 15, which represents a fatal obstacle. Semantic information 0001 represents a temporary obstacle. For the specific object, cost information and semantic information of another map layer are respectively expressed as: 1111 and 0010. Wherein, the cost value 1111 has a value of 15, which represents a fatal obstacle. The semantic information 0010 indicates a high expansion region (there is a correspondence between the high expansion region and 2). The mapping process can be seen in fig. 7, and the mapping result obtained is: the cost information and semantic information of the expansion layer are expressed as: 1111 and 0010. Wherein, the cost value 1111 has a value of 15, which represents a fatal obstacle. Semantic information 0010 indicates a high expansion region.

Because the expansion strategies of the region and the semantic information are different, in the process of barrier expansion (attenuation), not only the transmission with cost is needed, but also the transmission with expansion strategies is needed, the general principle is to keep the expansion strategies corresponding to high cost and high cost, and fig. 8 is the cost and expansion strategy process in the transmission process.

Referring to fig. 8, for a certain specific object, cost information and semantic information of a certain map layer are respectively expressed as: 1001 and 0001. Here, the cost value 1001 is 9, please refer to fig. 4, which illustrates that the obstacle space is inscribed. Semantic information 0001 represents a temporary obstacle. For the specific object, cost information and semantic information of another map layer are respectively expressed as: 0111 and 0010. The cost value 0111 is 7, please refer to fig. 4, which shows that the device is located in the external obstacle space, and the semantic information 0010 shows the high expansion strategy. The transfer procedure can be seen in fig. 8, and the mapping result obtained is: the cost information and semantic information of the expansion layer are expressed as: 1001 and 0001. Namely, the expansion strategy corresponding to high cost and high cost is transferred so as to ensure the driving safety of the robot.

For example, the semantic information is 16 kinds, and can be overlaid according to priority. If the vehicle priority is higher than that of the pedestrian, the vehicle semantics cover the pedestrian semantics when the point cloud of the same position is output, and the point expands according to the strategy of the vehicle when expanding. The expansion strategy is determined by the expansion distance and the attenuation coefficient, and when expansion is performed, the cost calculation is updated, and the sensor semantics are transferred until the maximum expansion range is exceeded. In the transmission process, higher cost should be reserved, and the semantics of the sensor with higher cost should be reserved.

In certain embodiments, the above method further comprises: if the expanded layer is a non-primary layer, multiple map layers and expanded layers are mapped to the primary layer to invoke the primary layer to plan the path.

In this embodiment, the main layer may be any one of the plurality of map layers. One or more of the multiple map layers may be mapped to an expansion layer for expansion processing. However, there may be a map layer that does not require expansion processing among the plurality of map layers.

Fig. 9 is a schematic diagram illustrating projection of an intumescent layer or the like onto a main layer according to an embodiment of the present application.

Referring to fig. 9, located at the bottom layer is a main layer, a plurality of map layers are present before the expansion layer, and a map layer is also present between the expansion layer and the main layer. The map layer before the expansion layer is mapped to the expansion layer for expansion processing. The map layer between the expansion layer and the main layer does not need to be mapped to the expansion layer, i.e. does not need to be subjected to expansion processing. To facilitate invoking the map, all map layers and expansion layers may be mapped to the main layer. This allows map data to be retrieved from the underlying layer.

In some embodiments, at least some of the plurality of map layers each have a corresponding cost map. By using a plurality of cost maps, elements with different confidence degrees are represented by a plurality of maps, expansion distinction aiming at different elements can be realized, and expansion results aiming at different elements are more visual and accurate. At the same time, this occupies more computing resources and memory.

Accordingly, mapping a plurality of map layers to an expanded layer includes: each of at least some of the map layers is mapped to an expanded layer of the cost map corresponding to the map layer, respectively. The mapping process may include overlay for cost value, overlay operations for semantic information, and so on.

Further, cost differences may be resolved based on the quality of the sensing capability or sensor data. For example, a map of high data quality may be given a higher weight to determine the cost value. For example, a plurality of sensors provided on a robot are limited by the viewing angle (affected by the installation position and direction) of each sensor, and the quality of data collected by some of the sensors is poor. Furthermore, one or more sensor perspectives may be blocked by other objects during operation and/or may be negatively affected by adverse weather and environmental conditions. At this time, a higher confidence may be given to the element (map layer) corresponding to the sensor that has high quality of the acquired data.

In some embodiments, a map of the main layer may be invoked for path planning. Static maps cannot update obstacle information on the map at any time. In a real-world environment, there are always various new obstacles that cannot be expected to appear in the current map, or old obstacles have now been removed from the environment map. In addition, since the default map is a black-white gray three-color map, namely, only the obstacle region, the free movement region and the unexplored region are marked, and the expansion processing is performed on different obstacles by the same expansion strategy. The robot performs path planning in such a map, which may result in an insufficiently safe planned path. According to the embodiment of the application, different expansion strategies are allocated to various different elements for the robot, so that different buffer areas are allocated to different obstacles, and the robot is safer when moving based on the map fetched from the main layer.

According to the method and the device for realizing the obstacle point expansion and attenuation of the map, different expansion strategies are used on the map, so that the map can be provided with the obstacle point expansion distance and attenuation coefficient according to different elements, the confidence or design difference between different elements can be embodied, and the collective perception strategy which is more credible and can be adapted to different environments can be provided.

FIG. 10 is a block diagram of a determination of an inflation strategy according to one embodiment of the present application.

Referring to fig. 10, the apparatus 1000 for determining an expansion policy may include: an element acquisition module 1010, an element mapping module 1020, and an expansion policy determination module 1030.

The element obtaining module 1010 is configured to obtain a plurality of elements, where each element has corresponding cost information and semantic information, and a corresponding relationship exists between the semantic information and the expansion policy.

The element mapping module 1020 is configured to map a plurality of elements to the expansion layer respectively, so as to obtain cost information and semantic information of each unit in the expansion layer.

The expansion policy determining module 1030 is configured to determine an expansion policy of each grid in the expansion layer based on the correspondence and the cost information and semantic information of each grid in the expansion layer.

In this embodiment of the present application, the apparatus 1000 further includes a construction module. The construction module is used for constructing map layers corresponding to each element, each map layer comprises a plurality of grids, and corresponding cost information and semantic information exist in each grid.

Correspondingly, the element mapping module 1020 is specifically configured to map a plurality of map layers to an expansion layer, so as to obtain cost information and semantic information of each grid in the expansion layer.

The specific manner in which the respective modules perform the operations in the apparatus of the above embodiments has been described in detail in the embodiments related to the method, and will not be described in detail herein.

In this embodiment, the cost map supports multiple expansion strategies. Different elements (such as prior map, area information, sensor category and semantic output) are reserved in the main layer except for superposition cost, expansion information with highest priority is reserved, and the main layer obstacle points are expanded according to the expansion information, so that the aim of classified expansion is fulfilled.

Another aspect of the present application also provides a mobile device.

Referring to fig. 11, a mobile device 1100 includes a mobile device body, a memory 1110, and a controller 1120. In addition, mobile device 1100 may also include a sensing device 1130.

The sensing device 1130 may be provided in one or more. For example, at least two sensing devices 1130 are respectively disposed at least two positions of the body for obtaining sensing information for the target viewing angle, such as for respectively acquiring target images with different viewing angles. For example, the sensing device 1130 is provided at the roof portion of the vehicle body, the front end of the vehicle body, or the like. For example, the sensing device 1130 may be at least one of a radar, a camera, or an interactive device. For example, the photographing device includes a body, a camera, and a photosensitive material. In addition, the system can also comprise aperture, shutter, ranging, framing, photometry, film conveying, counting, self-timer, focusing, zooming and the like. For example, the interaction device may be a touch screen or a mouse, etc., and the user may implement setting information such as a forbidden zone by operating the interaction device.

Memory 1110 may include various types of storage units, such as system memory, read Only Memory (ROM), and persistent storage. Wherein the ROM may store static data or instructions that are required by the processor or other modules of the computer. The persistent storage may be a readable and writable storage. The persistent storage may be a non-volatile memory device that does not lose stored instructions and data even after the computer is powered down. In some embodiments, the persistent storage device employs a mass storage device (e.g., magnetic or optical disk, flash memory) as the persistent storage device. In other embodiments, the persistent storage may be a removable storage device (e.g., diskette, optical drive). The system memory may be a read-write memory device or a volatile read-write memory device, such as dynamic random access memory. The system memory may store instructions and data that are required by some or all of the processors at runtime. Furthermore, memory 1110 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (e.g., DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), magnetic disks, and/or optical disks may also be employed. In some implementations, memory 1110 may include readable and/or writable removable storage devices such as Compact Discs (CDs), digital versatile discs (e.g., DVD-ROMs, dual-layer DVD-ROMs), blu-ray discs read only, super-density discs, flash memory cards (e.g., SD cards, min SD cards, micro-SD cards, etc.), magnetic floppy disks, and the like. The computer readable storage medium does not contain a carrier wave or an instantaneous electronic signal transmitted by wireless or wired transmission.

The controller 1120 may include: at least one processor. The processor may be a central processing unit (Central Processing Unit, abbreviated as CPU), other general purpose processor, digital signal processor (Digital Signal Processor, abbreviated as DSP), application specific integrated circuit (Application Specific Integrated Circuit, abbreviated as ASIC), field-programmable gate array (Field-Programmable Gate Array, abbreviated as FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The controller and/or processor is coupled to at least two perception devices for processing at least part of the image for the target viewing angle.

The memory 1110 has stored thereon executable code that, when processed by the processor, can cause the processor to perform some or all of the methods described above.

It is understood that, unless specifically stated otherwise, each functional unit/module in the embodiments of the present application may be integrated into one unit/module, or each unit/module may exist alone physically, or two or more units/modules may be integrated together. The integrated units/modules described above may be implemented either in hardware or in software program modules.

The integrated units/modules, if implemented in hardware, may be digital circuits, analog circuits, etc. Physical implementations of hardware structures include, but are not limited to, transistors, memristors, and the like. The artificial intelligence processor may be any suitable hardware processor, such as CPU, GPU, FPGA, DSP and ASIC, etc., unless otherwise specified. The Memory module may be any suitable magnetic or magneto-optical storage medium, such as resistive Random Access Memory (Resistive Random Access Memory, RRAM), dynamic Random Access Memory (Dynamic Random Access Memory, DRAM), static Random Access Memory (SRAM), enhanced dynamic Random Access Memory (Enhanced Dynamic Random Access Memory, EDRAM), high-Bandwidth Memory (HBM), or hybrid Memory cube (Hybrid Memory Cube, HMC), etc., unless otherwise indicated.

The integrated units/modules may be stored in a computer readable memory if implemented in the form of software program modules and sold or used as a stand-alone product. Based on such understanding, the technical solution of the present application may be embodied essentially or in part or all or part of the technical solution contributing to the prior art or in the form of a software product stored in a memory, including all or part of the steps of the method of the various embodiments of the present application if the instructions are dry to cause a computer device (which may be a personal computer, a server or a network device, etc.). And the aforementioned memory includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory RAM), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.

In one possible implementation, a board is also disclosed, which includes a memory device, an interface device, and a control device, and the processor; the processor is respectively connected with the storage device, the control device and the interface device; the storage device is used for storing data; the interface device is used for realizing data transmission between the processor and the external equipment; the control device is used for monitoring the state of the processor.

Another aspect of the present application also provides a controller (Control Unit). The controller includes at least one processor and a memory communicatively coupled to the at least one processor. Wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the method described above. The controller mainly comprises an instruction register, an instruction decoder and a control signal generator, and is mainly used for completing the translation of instructions, generating various control signals in and out of the chip and executing corresponding instructions.

The controller is a control center of the whole computer system and directs the computer to work in a coordinated manner, so that the computer is ensured to operate and process orderly according to the preset targets and steps. The controller takes out instructions from the memory one by one, analyzes what operation is specified by each instruction, the storage position of required data and the like, and then sends control signals to other parts of the computer according to the analysis result, and uniformly commands the whole computer to complete the operation specified by the instructions. The process of computer automatic operation is actually a process of automatically executing a program, and each instruction in the program is analyzed and executed by a controller, which is a main device controlled by the computer to implement the program.

Furthermore, the method according to the present application may also be implemented as a computer program or computer program product comprising computer program code instructions for performing part or all of the steps of the above-described method of the present application.

Alternatively, the present application may also be embodied as a computer-readable storage medium (or non-transitory machine-readable storage medium or machine-readable storage medium) having stored thereon executable code (or a computer program or computer instruction code) which, when executed by a processor of an electronic device (or a server, etc.), causes the processor to perform part or all of the steps of the above-described methods according to the present application.

The embodiments of the present application have been described above, the foregoing description is exemplary, not exhaustive, and not limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described. The terminology used herein was chosen in order to best explain the principles of the embodiments, the practical application, or the improvement of technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method of determining an expansion strategy, comprising:

obtaining a plurality of elements, wherein each element has corresponding cost information and semantic information, and a corresponding relation exists between the semantic information and an expansion strategy;

mapping a plurality of elements to an expansion layer respectively to obtain cost information and semantic information of each unit in the expansion layer;

and determining the expansion strategy of each grid in the expansion layer based on the corresponding relation and the cost information and semantic information of each grid in the expansion layer.

2. The method of claim 1, further comprising, after said obtaining a plurality of elements:

constructing map layers corresponding to each element, wherein each map layer comprises a plurality of grids, and corresponding cost information and semantic information exist in each grid;

mapping the plurality of elements to an expansion layer respectively to obtain cost information and semantic information of each unit in the expansion layer, wherein the method comprises the following steps:

and mapping a plurality of map layers to the expansion layer to obtain cost information and semantic information of each grid in the expansion layer.

3. The method of claim 2, wherein at least some of the map layers each have a corresponding cost map;

The mapping of the plurality of map layers to the expansion layer includes: each map layer of the at least partial map layers is mapped to an expanded layer of the cost map corresponding to the map layer.

4. The method of claim 2, wherein a common cost map exists for a plurality of said map layers;

the mapping of the plurality of map layers to the expansion layer includes: each map layer of the plurality of map layers is mapped to an expanded layer of the common cost map, respectively.

5. The method of claim 4, wherein the mapping each of the plurality of map layers to the inflated layer of the common cost map, respectively, comprises: for each target grid in the intumescent layer,

superposing a plurality of pieces of cost information of grids in a plurality of map layers corresponding to the target grid to obtain updated cost information of the target grid, and reserving semantic information with high priority in a plurality of pieces of candidate semantic information;

the cost information of the target grid is updated based on the updated cost information, and the semantic information of the target grid is updated based on the semantic information of the higher priority.

6. The method of claim 4, wherein the mapping each of the plurality of map layers to the inflated layer of the common cost map, respectively, comprises: for each of the map layers,

and updating the cost information and the semantic information corresponding to each grid in the expansion layer based on the cost information and the semantic information of each grid in the current map layer.

7. The method of claim 6, wherein updating the cost information and the semantic information corresponding to each grid in the expansion layer based on the cost information and the semantic information of each grid in the current map layer comprises: for each target grid in the intumescent layer,

superposing the cost information of the grids in the current map layer corresponding to the target grid with the cost information of the target grid, and updating the cost information of the target grid; and/or reserving semantic information with high priority in first semantic information and second semantic information, wherein the first semantic information is the semantic information of the target grid, and the second semantic information is the semantic information of the grid in the current map layer corresponding to the target grid.

8. The method of any one of claims 2 to 7, further comprising:

If the expanded layer is a non-primary layer, a plurality of the map layers and the expanded layer are mapped to a primary layer to invoke the primary layer to plan a path.

9. The method of any one of claims 1 to 7, wherein:

the elements include: at least one of a priori map, regional information, sensor class, semantic output; and/or

The cost information is characterized by numerical values, and the numerical values of the specific value range correspond to the specific collision risk probability range; and/or

The semantic information characterizes the artificial feature or the confidence of the corresponding position of the current grid; and/or

The parameters of the expansion strategy include at least one of an expansion radius and an attenuation coefficient.

10. The method of claim 9, wherein the cost information and the semantic information are characterized by a value having a predetermined byte length, a sum of the byte length of the cost information and the byte length of the semantic information being less than or equal to the predetermined byte length.

11. The method of claim 10, wherein the cost information is characterized using a first portion of bits in 1 bit and the cost information is characterized using a second portion of bits in the 1 bit.

12. An apparatus for determining an expansion strategy, comprising:

the element obtaining module is used for obtaining a plurality of elements, wherein each element has corresponding cost information and semantic information, and a corresponding relation exists between the semantic information and the expansion strategy;

the element mapping module is used for mapping a plurality of elements to an expansion layer respectively to obtain cost information and semantic information of each unit in the expansion layer;

and the expansion strategy determining module is used for determining the expansion strategy of each grid in the expansion layer based on the corresponding relation and the cost information and the semantic information of each grid in the expansion layer.

13. A mobile device, comprising:

a mobile device body;

the sensing device is arranged on the mobile device main body and is used for obtaining motion related information of the mobile device main body;

a controller, comprising at least one processor, coupled to the sensing device, for processing the motion-related information;

a memory having executable code stored thereon, which when executed by the processor causes the processor to perform the method of any of claims 1-11.

14. The mobile device of claim 13, wherein the sensing means comprises: at least one of a radar, a camera, or an interactive device.

15. A controller, comprising: at least one processor, and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the method of any one of claims 1-11.

16. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a program code, which is callable by a processor for executing the method according to any one of claims 1-11.