Robot topology map generation system, method, computer device and storage medium

Abstract

The present invention relates to the field of robot navigation, and discloses a robot topological map generation system, method, computer device and storage medium. The system includes a memory, a processor and a computer-readable instruction stored in the memory and executable on the processor. When the processor executes the computer-readable instruction, the following steps are implemented: obtaining the action trajectory of the robot in the mapping scene and the scene map of the mapping scene; filtering the action trajectory to generate a uniform point trajectory; mapping the uniform point trajectory to the scene map to generate a mapping scene map; inputting the mapping scene map into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model. The present invention greatly improves the efficiency of drawing topological maps and reduces human errors in topological maps.

Description

Technical Field

[0001] The present invention relates to the field of robot navigation, and in particular to a robot topology map generation system, method, computer equipment and storage medium. Background Art

[0002] In the process of robot automatic navigation, it usually relies on pre-drawn topological maps. Figure 1 However, for complex scenes, the topological map drawing process is cumbersome, time-consuming and labor-intensive, and there is a situation where the topological map does not match the actual scene, which affects the normal operation of the robot. Summary of the invention

[0003] Based on this, it is necessary to provide a robot topology map generation system, method, computer device and storage medium to address the above technical problems, so as to improve the efficiency of topology map drawing and reduce human errors in topology map.

[0004] A robot topology map generation system, comprising:

[0005] Obtaining the movement trajectory of the robot in the mapping scene and the scene map of the mapping scene;

[0006] Performing filtering on the motion trajectory to generate a homogenized point trajectory;

[0007] Mapping the homogenized point trajectory to the scene map to generate a mapped scene map;

[0008] The mapping scene map is input into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model.

[0009] A method for generating a robot topology map, comprising:

[0010] Obtaining the movement trajectory of the robot in the mapping scene and the scene map of the mapping scene;

[0011] Performing filtering on the motion trajectory to generate a homogenized point trajectory;

[0012] Mapping the homogenized point trajectory to the scene map to generate a mapped scene map;

[0013] The mapping scene map is input into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model.

[0014] A computer device comprises a memory, a processor and computer-readable instructions stored in the memory and executable on the processor, wherein the processor implements the robot topology map generating method when executing the computer-readable instructions.

[0015] One or more readable storage media storing computer-readable instructions, wherein when the computer-readable instructions are executed by one or more processors, the one or more processors execute the robot topology map generation method as described above.

[0016] The above-mentioned robot topology map generation method, device, computer equipment and storage medium obtain the action trajectory of the robot in the mapping scene, and the scene map of the mapping scene. Here, the difficulty of obtaining the action trajectory and the scene map is low, which is conducive to reducing the cost of drawing the topology map. The action trajectory is filtered to generate a uniform point trajectory to standardize the action trajectory and improve the accuracy of image processing. The uniform point trajectory is mapped to the scene map to generate a mapping scene map. Here, the mapping scene map integrates the trajectory information and the scene information. The mapping scene map is input into the image conversion model to obtain the topology map of the mapping scene output by the image conversion model. Here, the image conversion model directly converts the mapping scene map, which greatly improves the efficiency of generating the topology map. The present invention only needs to obtain the action trajectory and the scene map to automatically generate the topology map, which greatly improves the efficiency of drawing the topology map and reduces the human error of the topology map. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0018] Figure 1 is a schematic diagram of a process of a robot topology map generation system in one embodiment of the present invention;

[0019] Figure 2 is a mapping scene map containing a homogenized point trajectory after filtering in one embodiment of the present invention;

[0020] Figure 3 is a topological map after inversion processing in one embodiment of the present invention;

[0021] Figure 4 is a schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] like Figure 1 As shown, the robot topology map generation system includes a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, the following steps S10-S40 are implemented.

[0024] S10: Obtain the movement trajectory of the robot in the mapping scene and the scene map of the mapping scene.

[0025] Understandably, the mapping scene can be a robot's working scene, such as a restaurant, a hotel, a hospital, etc. The action trajectory can be the walking trajectory of the robot in the mapping scene. The action trajectory includes point trajectories of several robots. In one example, these point trajectories can be collected at preset time intervals. The preset time interval can be set according to actual needs. For example, the preset time interval can be 1s to 10s.

[0026] The scene map can be a pre-drawn flat map of the mapping scene, or a map drawn using depth image data. Generally, the scene map is a two-dimensional map.

[0027] S20: Filter the action trajectory to generate a homogenized point trajectory.

[0028] Understandably, in the action trajectory, the time intervals of adjacent trajectory points are equal, but there is a certain difference in the spacing between adjacent trajectory points. This spacing is related to the movement speed of the robot within the time interval. The action trajectory needs to be filtered to reduce the difference in spacing and obtain a homogenized point trajectory. A homogenized point trajectory means that the spacing between adjacent trajectory points is within a distance range to ensure that the spacing is uniform and not too large or too small.

[0029] S30: Map the homogenized point trajectory to the scene map to generate a mapped scene map.

[0030] Understandably, the homogenized point trajectory can be mapped to the scene map according to a preset mapping rule (such as a scale factor) to generate a mapped scene map. Figure 2 As shown, Figure 2 is the mapping scene map containing the homogenized point trajectory after filtering. Figure 2 The linear curve in is the homogenization point trajectory.

[0031] S40: Input the mapping scene map into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model.

[0032] It can be understood that the image conversion model is a model built based on a preset image processing algorithm. The image conversion model can realize image-to-image translation. Here, the image processing algorithm can be a deep convolutional neural network (CNNs), a conditional generative adversarial network (CGAN), etc. Figure 3 As shown, Figure 3 It is a topological map after inversion processing.

[0033] In steps S10-S40, the action trajectory of the robot in the mapping scene and the scene map of the mapping scene are obtained. Here, the difficulty of obtaining the action trajectory and the scene map is low, which is conducive to reducing the cost of drawing the topological map. The action trajectory is filtered to generate a uniform point trajectory to standardize the action trajectory and improve the accuracy of image processing. The uniform point trajectory is mapped to the scene map to generate a mapping scene map. Here, the mapping scene map integrates the trajectory information and the scene information. The mapping scene map is input into the image conversion model to obtain the topological map of the mapping scene output by the image conversion model. Here, the image conversion model directly converts the mapping scene map, which greatly improves the efficiency of generating the topological map. In this embodiment, only the action trajectory and the scene map need to be obtained to automatically generate the topological map, which greatly improves the efficiency of drawing the topological map and reduces human errors in the topological map.

[0034] Optionally, step S10, i.e., obtaining the action trajectory of the robot in the mapping scene and the scene map of the mapping scene, includes:

[0035] S101, acquiring depth image data collected when the robot passes through the motion track, wherein the depth image data is collected by a depth camera and / or a laser radar;

[0036] S102, stitching the depth image data according to the position of the robot in the motion trajectory to generate a three-dimensional point cloud of the mapping scene;

[0037] S103: Perform two-dimensional projection on the three-dimensional point cloud to generate the scene map.

[0038] Understandably, the depth image data (RGBD) may be image data acquired by a depth camera and / or a laser radar mounted on the robot. The depth camera may collect depth image data at different track points at a preset time frequency. By splicing these depth image data based on the positions of each track point, a three-dimensional point cloud of the mapping scene may be generated. The three-dimensional point cloud is projected two-dimensionally to generate a two-dimensional scene map. Here, the three-dimensional point cloud is projected onto the ground plane (which may be a horizontal plane) of the mapping scene to generate a two-dimensional scene map.

[0039] Optionally, step S20, namely, filtering the action trajectory to generate a homogenized point trajectory, includes:

[0040] S201, interpolating the action trajectory, and then performing median filtering and mean smoothing processing to generate the homogenized point trajectory; in the homogenized point trajectory, the distance between adjacent trajectory points is greater than a first threshold and less than a second threshold.

[0041] Understandably, the action trajectory can be interpolated to increase the number of trajectory points so that the spacing between all adjacent trajectory points is less than the first threshold. Then, the trajectory points with too small spacing are removed by median filtering, and then the spacing between adjacent trajectory points is roughly equal by mean smoothing, thereby obtaining a homogenized point trajectory. In the homogenized point trajectory, the spacing between adjacent trajectory points is greater than the first threshold and less than the second threshold.

[0042] Optionally, the first threshold includes 0.25m, and the second threshold includes 1m.

[0043] It is understandable that the first threshold and the second threshold can be set according to actual needs. In one example, the first threshold includes 0.25m, and the second threshold includes 1m.

[0044] Optionally, step S30, i.e. mapping the homogenized point trajectory to the scene map to generate a mapped scene map, includes:

[0045] S301, mapping the homogenized point trajectory to the scene map to generate an initial mapping scene map;

[0046] S302: Perform image filtering on the initial mapping scene map to generate the mapping scene map.

[0047] Understandably, in step S301, the homogenized point trajectory can be mapped to the scene map to generate an initial mapping scene map. When mapping, the coordinates of the trajectory points in the homogenized point trajectory can be proportionally enlarged or reduced. For example, the scale factor is 0.05, the coordinates of the trajectory point are (100, 100), and after being mapped to the scene map, the trajectory mapping coordinates corresponding to the trajectory point are (5, 5).

[0048] In step S302, the initial mapping scene map may be subjected to image median filtering and image mean filtering to generate a mapping scene map. Image median filtering and image mean filtering may filter out noise points in the initial mapping scene map and optimize the image quality of the mapping scene map.

[0049] Optionally, step S301, i.e. mapping the homogenized point trajectory to the scene map to generate an initial mapping scene map, includes:

[0050] S3011, obtaining a scale factor of the homogenization point trajectory;

[0051] S3012, converting the point coordinates of the homogenized point trajectory into mapping coordinates according to the scale factor;

[0052] S3013: adding track points to the scene map according to the mapping coordinates to generate the track mapping image.

[0053] Understandably, the scale factor is equivalent to the scale of the map, which can enlarge or reduce the coordinates of the trajectory point. The scale factor changes as the ratio between the homogenized point trajectory and the scene map changes. Through step S3011, the scale factor of the homogenized point trajectory can be obtained. In an example, the value of the scale factor is 0.05.

[0054] In step S3012, the point coordinates of the homogenized point trajectory may be converted into mapping coordinates according to the scale factor. In one example, the mapping coordinates are the product of the point coordinates of the homogenized point trajectory and the inverse of the scale factor, which is expressed by the formula:

[0055]

[0056] Among them, P1(X1, Y1) is the mapping coordinate, P(X, Y) is the point coordinate of the homogenized point trajectory, and scale is the scale factor.

[0057] In step S3013, track points can be added to the scene map according to the mapping coordinates to generate a track mapping image. The track mapping image includes both the obstacle information in the original scene map and the track data of the robot.

[0058] Optionally, before step S40, that is, before inputting the mapping scene map into the image conversion model and obtaining the topological map of the mapping scene output by the image conversion model, the method further includes:

[0059] S41, acquiring training data, wherein the training data includes a training map and a target trajectory image;

[0060] S42, using the training data to train the image processing model so that the image processing model learns the association relationship between the training map and the target trajectory image;

[0061] S43, determining the trained image processing model as the image conversion model, wherein the image conversion model is used to convert the mapping scene map into a topological map according to the trained association relationship, and the conversion accuracy is not lower than a preset threshold.

[0062] Understandably, in step S41, a number of training data can be obtained, each training data includes a training map and a target trajectory image. Among them, the training map is a scene map containing the robot's motion trajectory; the target trajectory image is a topological map corresponding to the training map. Here, the topological map can be a manually drawn map. Based on actual needs, training maps for a variety of different scenarios can be selected. These scenes can be restaurants, hotels, hospitals, schools, etc. The number of training maps for each scene is not less than a preset number threshold. The preset number threshold can be set according to the actual needs of the image processing model. In other words, different image processing models may have different corresponding preset number thresholds. In some examples, the training map can be annotated to generate scene labels.

[0063] In step S42, a suitable image processing model, such as a model based on a deep convolutional neural network, a model based on a conditional adversarial network, etc., can be selected, and the training data can be input into the image processing model for training. During the training process, the model can learn the association between the training map and the target trajectory image, and the association can reflect the drawing characteristics of the topological path in the target trajectory image, such as considering the obstacle information around the path, the distribution of the path nodes, the building plane layout, etc.

[0064] In step S43, when the image processing model meets the preset convergence condition, the image processing model training is completed. The convergence condition can be set according to the algorithm involved in the image processing model and the actual use needs. In some examples, the training map can be converted into a trajectory image by the image processing model, and then the difference between the trajectory image and the target trajectory image is calculated, and the parameters of the image processing model are adjusted according to the difference. When the conversion accuracy of the image processing model is not lower than the preset threshold, the training can be terminated. The preset threshold can be set according to actual needs, such as 85%, 90%, 95%, etc. The trained image processing model is determined as the image conversion model.

[0065] In one embodiment, a method for generating a robot topology map is provided, and the robot topology map generating system corresponds one-to-one to the robot topology map generating system in the above embodiment. Figure 1As shown, the robot topology map generation method provided in this embodiment includes the following steps:

[0066] S10, obtaining a movement trajectory of the robot in the mapping scene and a scene map of the mapping scene;

[0067] S20, filtering the action trajectory to generate a homogenized point trajectory;

[0068] S30, mapping the homogenized point trajectory to the scene map to generate a mapped scene map;

[0069] S40: Input the mapping scene map into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model.

[0070] Optionally, step S10, i.e., obtaining the action trajectory of the robot in the mapping scene and the scene map of the mapping scene, includes:

[0071] S101, acquiring depth image data collected when the robot passes through the motion track, wherein the depth image data is collected by a depth camera and / or a laser radar;

[0072] S102, stitching the depth image data according to the position of the robot in the motion trajectory to generate a three-dimensional point cloud of the mapping scene;

[0073] S103: Perform two-dimensional projection on the three-dimensional point cloud to generate the scene map.

[0074] Optionally, in step S20, filtering the action trajectory to generate a homogenized point trajectory includes:

[0075] S201, interpolating the action trajectory, and then performing median filtering and mean smoothing processing to generate the homogenized point trajectory, in which the distance between adjacent trajectory points is greater than a first threshold and less than a second threshold.

[0076] Optionally, the first threshold includes 0.25m, and the second threshold includes 1m.

[0077] Optionally, in step S30, mapping the homogenized point trajectory to the scene map to generate a mapped scene map includes:

[0078] S301, mapping the homogenized point trajectory to the scene map to generate an initial mapping scene map;

[0079] S302: Perform image filtering on the initial mapping scene map to generate the mapping scene map.

[0080] Optionally, step S301, i.e. mapping the homogenized point trajectory to the scene map to generate an initial mapping scene map, includes:

[0081] S3011, obtaining a scale factor of the homogenization point trajectory;

[0082] S3012, converting the point coordinates of the homogenized point trajectory into mapping coordinates according to the scale factor;

[0083] S3013. Add track points to the scene map according to the mapping coordinates to generate the initial mapping scene map.

[0084] Optionally, before step S40, that is, before inputting the mapping scene map into the image conversion model and obtaining the trajectory image output by the image conversion model, the method further includes:

[0085] S41, acquiring training data, wherein the training data includes a training map and a target trajectory image;

[0086] S42, using the training data to train the image processing model so that the image processing model learns the association relationship between the training map and the target trajectory image;

[0087] S43, determining the trained image processing model as the image conversion model, wherein the image conversion model is used to convert the mapping scene map into a topological map according to the trained association relationship, and the conversion accuracy is not lower than a preset threshold.

[0088] In an optional embodiment, the training of the image processing model using the training data may be offline training or online training, which is not limited here.

[0089] The specific limitations of the robot topology map generation method can be found in the above limitations of the robot topology map generation system, which will not be repeated here. It should be understood that the size of the sequence number of each step in the above embodiment does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0090] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a readable storage medium and an internal memory. The readable storage medium stores an operating system, computer-readable instructions and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the readable storage medium. The database of the computer device is used to store data involved in the robot topology map generation method. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer-readable instructions are executed by the processor, a robot topology map generation method is implemented. The readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0091] In one embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the processor executes the computer-readable instructions to implement the following steps:

[0092] Obtaining the movement trajectory of the robot in the mapping scene and the scene map of the mapping scene;

[0093] Performing filtering processing on the action trajectory to generate a homogenized point trajectory;

[0094] Mapping the homogenized point trajectory to the scene map to generate a mapped scene map;

[0095] The mapping scene map is input into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model.

[0096] In one embodiment, one or more computer-readable storage media storing computer-readable instructions are provided. The readable storage media provided in this embodiment include non-volatile readable storage media and volatile readable storage media. The readable storage media store computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the following steps are implemented:

[0097] Obtaining the movement trajectory of the robot in the mapping scene and the scene map of the mapping scene;

[0098] Performing filtering processing on the action trajectory to generate a homogenized point trajectory;

[0099] Mapping the homogenized point trajectory to the scene map to generate a mapped scene map;

[0100] The mapping scene map is input into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model.

[0101] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through computer-readable instructions, and the computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When the computer-readable instructions are executed, they may include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0102] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0103] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A robot topology map generation system, comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, characterized in that: The processor is used to execute the computer-readable instructions to implement the following steps: Obtaining the movement trajectory of the robot in the mapping scene and the scene map of the mapping scene; Performing filtering on the motion trajectory to generate a homogenized point trajectory; Mapping the homogenized point trajectory to the scene map to generate a mapped scene map; Inputting the mapping scene map into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model; When the processor executes the computer-readable instructions, the processor also implements the following steps: Acquire training data, wherein the training data includes a training map and a target trajectory image, wherein the training map is a scene map including a robot motion trajectory, and the target trajectory image is a topological map corresponding to the training map; Using the training data to train an image processing model so that the image processing model learns the association relationship between the training map and the target trajectory image; The trained image processing model is determined as the image conversion model, and the image conversion model is used to convert the mapping scene map into a topological map according to the trained association relationship, and the conversion accuracy is not lower than a preset threshold.

2. The robot topology map generation system according to claim 1, characterized in that: The obtaining of the movement trajectory of the robot in the mapping scene and the scene map of the mapping scene includes: Acquiring depth image data collected when the robot passes through the motion track; splicing the depth image data according to the position of the robot in the motion trajectory to generate a three-dimensional point cloud of the mapping scene; The three-dimensional point cloud is two-dimensionally projected to generate the scene map.

3. The robot topology map generation system according to claim 1, characterized in that: The filtering process of the motion trajectory to generate a homogenized point trajectory includes: The action trajectory is interpolated, and then median filtering and mean smoothing are performed to generate the homogenized point trajectory, in which the distance between adjacent trajectory points is greater than a first threshold and less than a second threshold.

4. The robot topology map generation system according to claim 3, characterized in that: The first threshold is 0.25 m inclusive, and the second threshold is 1 m inclusive.

5. The robot topology map generation system according to claim 1, characterized in that: Mapping the homogenized point trajectory to the scene map to generate a mapped scene map includes: Mapping the homogenized point trajectory to the scene map to generate an initial mapping scene map; Perform image filtering on the initial mapping scene map to generate the mapping scene map.

6. The robot topology map generation system according to claim 5, characterized in that: Mapping the homogenized point trajectory to the scene map to generate an initial mapping scene map includes: Obtaining a scale factor of the homogenization point trajectory; converting the point coordinates of the homogenized point trajectory into mapping coordinates according to the scale factor; Track points are added to the scene map according to the mapping coordinates to generate the initial mapping scene map.

7. A method for generating a robot topology map, characterized in that: include: Obtaining the movement trajectory of the robot in the mapping scene and the scene map of the mapping scene; Performing filtering on the motion trajectory to generate a homogenized point trajectory; Mapping the homogenized point trajectory to the scene map to generate a mapped scene map; Inputting the mapping scene map into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model; Before inputting the mapping scene map into an image conversion model to obtain a topological map of the mapping scene output by the image conversion model, the method further includes: Acquire training data, wherein the training data includes a training map and a target trajectory image, wherein the training map is a scene map including a robot motion trajectory, and the target trajectory image is a topological map corresponding to the training map; Using the training data to train an image processing model so that the image processing model learns the association relationship between the training map and the target trajectory image; The trained image processing model is determined as the image conversion model, and the image conversion model is used to convert the mapping scene map into a topological map according to the trained association relationship, and the conversion accuracy is not lower than a preset threshold.

8. A computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, characterized in that: When the processor executes the computer-readable instructions, the robot topology map generation method as claimed in claim 7 is implemented.

9. One or more readable storage media storing computer-readable instructions, wherein when the computer-readable instructions are executed by one or more processors, the one or more processors execute the robot topology map generation method as described in claim 7.

Images