Autonomous exploration method based on structured environment and related equipment

Abstract

The invention relates to an autonomous exploration method based on a structured environment and related equipment, and the method comprises the steps: obtaining multi-frame first local point cloud data collected by a laser radar at a historical moment and attitude information recorded by an IMU sensor at a corresponding historical moment, splicing multiple frames of first local point cloud data by using the attitude information corresponding to different historical moments to obtain a global point cloud; plane fitting is carried out on the global point cloud to obtain point cloud plane features, feature division is carried out on the point cloud plane features through preset geometric features to obtain structured point clouds, and the structured point clouds comprise different room plane point clouds and corridor plane point clouds; and performing environment segmentation and topology construction by using the structured point cloud to obtain a global topological graph, and performing global path planning based on the global topological graph to obtain a target path. The method can better adapt to large-scale, complex and dynamically changing scenes, and realizes efficient and accurate autonomous exploration.

Description

Technical Field

[0001] This application belongs to the field of computer vision technology and relates to an autonomous exploration method and related devices based on a structured environment. Background Art

[0002] With the rapid development of robotics in key areas such as automation, intelligent navigation, and autonomous exploration, the autonomous exploration ability has become one of the core elements of robotics. In practical applications, robots often need to independently complete exploration, positioning, and navigation tasks in unknown or dynamically changing environments. Currently, most autonomous exploration methods rely on environmental perception, path planning, and decision-making systems. They collect environmental information through sensors such as lidar (LiDAR) and combine path planning algorithms to achieve navigation, thereby helping robots perceive the surrounding environment and efficiently plan the optimal path to complete tasks.

[0003] Traditional autonomous exploration methods are mainly based on the frontier-based exploration algorithm, that is, the robot perceives the boundaries of unknown areas and selects frontier points as the next exploration goal. These methods perform well in small-scale or static environments, but in large-scale and complex environments, it is difficult for the robot to fully understand the structural information of the environment, resulting in low exploration efficiency. To improve exploration efficiency, some existing algorithms introduce graph-based path planning methods, which optimize path planning by constructing a topological graph of the environment to reduce redundant exploration. However, these methods still have limitations in dynamic environments and are difficult to adapt to the rapid changes in the environment in real time.

[0004] In recent years, to address the limitations of existing methods, researchers have proposed a series of exploration strategies that focus on integrating the structured information of the environment, topological graph construction, and hierarchical exploration planning to improve the exploration efficiency and accuracy of robots in complex dynamic environments. Nevertheless, the application of existing methods in complex and dynamic environments still faces many challenges: on the one hand, existing methods fail to fully utilize the structured information in the environment, such as features like rooms and corridors, resulting in low exploration efficiency in large-scale complex environments; on the other hand, many methods have a high computational complexity in large-scale environments, are prone to generating redundant paths and repeated explorations, and are difficult to efficiently handle complex scenarios; in addition, existing methods have poor adaptability to dynamic environmental changes and cannot adjust the planning strategy in real time, thus affecting the exploration accuracy and efficiency. In view of the above situation, there is an urgent need for a new exploration method to solve the problems of large-scale, complex, and dynamic environments.

[0005] The content in the background art section is only the publicly known technology and does not necessarily represent the prior art in this field. Summary of the Invention

[0006] The purpose of this application is to solve the deficiencies in the prior art to at least a certain extent. The first aspect of this application provides an active exploration method based on a structured environment, including: obtaining multiple frames of first local point cloud data collected by a lidar at historical moments and the attitude information recorded by an IMU sensor at the corresponding historical moments, and splicing the multiple frames of first local point cloud data using the attitude information corresponding to different historical moments to obtain a global point cloud; performing plane fitting on the global point cloud to obtain point cloud plane features, and performing feature division on the point cloud plane features through preset geometric features to obtain structured point clouds, where the structured point clouds include different room plane point clouds and corridor plane point clouds; using the structured point clouds for environment segmentation and topology construction to obtain a global topology map and performing global path planning based on the global topology map to obtain a target path.

[0007] The second aspect of this application provides an active exploration device based on a structured environment. The device includes: an acquisition unit for obtaining multiple frames of first local point cloud data collected by a lidar at historical moments and the attitude information recorded by an IMU sensor at the corresponding historical moments, and splicing the multiple frames of first local point cloud data using the attitude information corresponding to different historical moments to obtain a global point cloud; a division unit for performing plane fitting on the global point cloud to obtain point cloud plane features, and performing feature division on the point cloud plane features through preset geometric features to obtain structured point clouds, where the structured point clouds include different room plane point clouds and corridor plane point clouds; a planning unit for using the structured point clouds for environment segmentation and topology construction to obtain a global topology map and performing global path planning based on the global topology map to obtain a target path.

[0008] The third aspect of this application provides an active exploration system based on a structured environment, where: the lidar is used to scan the indoor environment including corridors and multiple rooms where the exploration system is located at different moments to obtain local point cloud data from different perspectives; the IMU sensor is used to record the attitude information of the exploration system while the lidar collects point cloud data; the processing platform is used to process the local point cloud data and attitude information at different moments using the above active exploration method to obtain a target path.

[0009] The fourth aspect of this application provides a computer-readable storage medium storing a computer program, and the computer program is executed by a processor to implement the active exploration method based on a structured environment described in the first aspect.

[0010] The fifth aspect of this application provides a computer program product. When the computer program code or instructions are executed on a computer, the computer is caused to execute the active exploration method based on a structured environment described in the first aspect.

[0011] As can be seen from the above embodiments of the present application, the present application constructs a structured topology graph to accurately extract environmental structure information such as rooms and corridors, and performs global path planning based on the environmental structure information. The global path is optimized through a local acquisition vision dynamic adjustment strategy, which can not only effectively reduce the computational complexity, maintain high stability and accuracy in a dynamic environment, but also better adapt to large-scale, complex and dynamically changing scenarios, realizing efficient and accurate autonomous exploration. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those skilled in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0013] Figure 1 It is a schematic structural diagram of a positioning system provided by an embodiment of the present application; Figure 2 It is a schematic flow diagram of an active exploration method based on a structured environment provided by an embodiment of the present application; Figure 3 It is a schematic block diagram of an active exploration device based on a structured environment provided by an embodiment of the present application; Figure 4 It is a schematic block diagram of a computer device provided by an embodiment of the present application; Figure 5 It is a schematic block diagram of a computer-readable storage medium provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] The following details the embodiments of the present application. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions from beginning to end. The embodiments described below with reference to the drawings are exemplary and are intended to explain the present application, but should not be construed as limiting the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0015] Figure 1It is a schematic architecture diagram of an active exploration system 100 based on a structured environment proposed by an embodiment of the present application. The exploration system 100 includes a perception sensor 110, an IMU sensor 120, and a processing platform 130. Among them, the perception sensor 110 includes a lidar, and the lidar is used to scan the indoor environment where the exploration system 100 is located, including corridors and multiple rooms, at different times to obtain local point cloud data from different perspectives; the IMU sensor 120 is used to record the pose information of the exploration system while the lidar collects point cloud data; the processing platform 130 is used to perform environment segmentation, structured mapping, global planning, etc. on the local point cloud data and pose information at different times through the exploration method based on the structured environment provided by one or more embodiments of the present application to obtain a target path.

[0016] Furthermore, in a low-light or feature-scarce environment, effective information cannot be extracted only relying on the point cloud data collected by the lidar. Therefore, in the present application, the perception sensor 110 further includes a camera, which can be an infrared camera, a grayscale camera, a color camera, a depth camera, etc., and is used to collect image data of the surrounding environment of the exploration system 100 to assist in environment segmentation and structured mapping, so as to fuse multiple sensor data to improve the performance in a low-light or feature-scarce environment.

[0017] Among them, the camera is preferably an RGB-D camera. The RGB-D camera includes an RGB sensor and a depth sensor. Due to factors such as the depth sensor being unaffected by low light and the RGB sensor being able to capture rich texture information, it can capture effective feature information in a low-light or feature-scarce environment, thereby assisting in the environment segmentation and structured mapping of local point cloud data to generate a more accurate exploration path.

[0018] In some embodiments, the processing platform 120 may include a processor and a memory. Among them, the processor is a circuit with signal processing capabilities, and the memory is used to store program instructions corresponding to the exploration method based on the structured environment provided by one or more embodiments of the present application. Some or all of the processors in the processor can call the instructions in the memory to perform environment segmentation and structured mapping on the point cloud data collected by the lidar according to the exploration method provided by the present application and the pose information recorded by the IMU sensor to obtain an exploration path.

[0019] Specifically, the processor can be divided into a structured hierarchical exploration and planning module and a local planner according to its implemented functions. The local planner is used to construct a global topological structure of the environment where the exploration system is located based on the local point cloud data at different times to obtain a global topological map, and perform global path planning based on the global topological map to obtain a target path. The structured hierarchical exploration and planning module is used to dynamically adjust the acquisition field of view of the lidar, so as to optimize the target path by combining the local point cloud data collected by the lidar with the adjusted acquisition field of view to generate a collision-free and smooth exploration path.

[0020] In one embodiment, the processor can be a circuit with the ability to read and execute instructions, such as a central processing unit (CPU), a microprocessor, a graphics processing unit (GPU) (which can be understood as a type of microprocessor), or a digital signal processor (DSP), etc. In another implementation, the processor can implement certain functions through the logical relationship of a hardware circuit, and the logical relationship of the hardware circuit is fixed or can be reconfigured. For example, the processor is a hardware circuit implemented by an application-specific integrated circuit (ASIC) or a programmable logic device (PLD), such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document to implement the configuration of the hardware circuit can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units. In addition, the processor can also be a hardware circuit designed for artificial intelligence, which can be understood as a type of ASIC, such as a neural network processing unit (NPU), a tensor processing unit (TPU), a deep learning processing unit (DPU), etc.

[0021] In addition, the exploration system 100 of the present application can be applied to intelligent devices, which can include: road vehicles, water vehicles, air vehicles, industrial equipment, agricultural equipment, wearable devices, mobile robots, or entertainment devices, etc. For example, the intelligent device can be a vehicle, which is a vehicle in a broad sense and can be a transportation vehicle (such as a commercial vehicle, a passenger vehicle, a motorcycle, a flying car, a train, etc.), an industrial vehicle (such as a forklift, a trailer, a tractor, etc.), an engineering vehicle (such as an excavator, a bulldozer, a crane, etc.), an agricultural equipment (such as a lawn mower, a harvester, etc.), a recreational device, a toy vehicle, etc. The embodiments of the present application do not specifically limit the type of the vehicle.

[0022] Figure 2 is a schematic flow chart of an active exploration method based on a structured environment provided by an embodiment of the present application. This method is applied in the exploration system 100 as shown in Figure 1 and specifically includes: S210: Obtain multiple frames of first local point cloud data collected by the lidar at historical moments and the attitude information recorded by the IMU sensor at the corresponding historical moments, and splice the multiple frames of first local point cloud data using the attitude information corresponding to different historical moments to obtain a global point cloud.

[0023] Specifically, when the lidar is used to scan the indoor environment including corridors and multiple rooms where the exploration system 100 is located to obtain multiple frames of local point cloud data from different perspectives, the IMU sensor is used to record the attitude information of the exploration system 100 while the lidar is working. Among them, the lidar and the IMU sensor share the system timestamp of the exploration system 100, and the relative positions of the lidar and the IMU sensor are fixed during the working process. In view of the fact that the relative positions of the lidar and the IMU sensor are fixed during the working process, so that at the same timestamp, at least one frame of local point cloud data collected by the lidar has a one-to-one correspondence with the attitude information collected by the IMU sensor.

[0024] In one embodiment, the first relative transformation relationship between the local point cloud data from different perspectives is obtained through the Iterative Closest Point (ICP) algorithm, the attitude deviation of the first relative transformation relationship is corrected using the attitude information recorded by the IMU sensor to obtain a second relative transformation relationship, and the local point cloud data from different perspectives are spliced based on the second relative transformation relationship to obtain a global point cloud. Furthermore, considering the large amount of multi-frame local point cloud data and the possible existence of outliers, directly processing the multi-frame local point cloud will lead to large computational complexity and low accuracy. Therefore, before stitching the multi-frame local point cloud, necessary preprocessing can be performed on the multi-frame local point cloud, such as filtering to remove noise (such as outliers), downsampling to reduce the data volume, etc., to retain the main structural features and efficiently obtain a more accurate global point cloud.

[0025] S220: Perform plane fitting and feature division on the global point cloud to obtain a structured point cloud, use the structured point cloud for environment segmentation and topology construction to obtain a global topology map, and perform global path planning based on the global topology map to obtain a target path.

[0026] In some embodiments, performing plane fitting and environment segmentation on the global point cloud to obtain a structured point cloud includes: performing plane fitting on the global point cloud to obtain point cloud plane features, and performing feature division on the point cloud plane features through preset geometric features to obtain a structured point cloud, where the structured point cloud includes different room plane point clouds and corridor plane point clouds.

[0027] In one embodiment, performing plane fitting on the global point cloud to obtain point cloud plane features includes: randomly selecting three points in the global point cloud as sample points to construct a first plane and fitting the first plane equation of the first plane, calculating the distances from the remaining points in the global point cloud to the plane based on the first plane equation, and if the distance is less than a preset threshold, regarding the corresponding points as inliers to record the first inlier quantity and the first inlier set; if the first inlier quantity recorded at the current moment is greater than the first inlier quantity recorded at the previous moment, updating the plane equation based on the first inlier set corresponding to the current moment, and repeating the steps of "calculating the distances from the remaining points in the global point cloud to the first plane based on the first plane equation, and regarding the points with a distance less than the preset threshold as inliers to record the first inlier quantity and the first inlier set" until all points in the global point cloud are traversed to obtain the second inlier set corresponding to the maximum inlier quantity, and using it as the inlier set of the best-fitting plane; randomly selecting three inliers in the second inlier set to construct the initial plane equation of the second plane, and determining the plane parameters of the initial plane equation by minimizing the distances from each inlier in the second inlier set to the second plane to obtain the second plane equation, and the plane corresponding to the second plane equation is the best-fitting plane, which is the point cloud plane feature.

[0028] Preferably, the fitted first plane equation, initial plane equation, and / or second plane equation are specifically expressed as:

[0029] where a, b, and c are the normal vectors of the plane, and d is the offset of the plane.

[0030] In one embodiment, the point cloud plane features are divided by preset geometric features to obtain a structured point cloud, including: using the preset geometric features that the room has multiple parallel walls, the ground, and the corridor has a long and straight geometric shape to divide the point cloud plane features into point clouds of different attribute types to obtain a structured point cloud including room plane point clouds and corridor plane point clouds.

[0031] It should be noted that, in addition to the above method, it is also possible to obtain a structured point cloud including room plane point clouds and corridor plane point clouds through an environment semantic segmentation method based on deep learning to achieve automated environment understanding. This application does not limit this here. In addition, after obtaining the room plane point clouds and corridor plane point clouds above, the room plane point clouds and corridor plane point clouds can also be used as semantic landmark points, and the corresponding local point clouds obtained by using them and lidar can be used together for loop detection and global optimization of the global point cloud, which can further improve the consistency and accuracy of the global point cloud.

[0032] This embodiment can efficiently and accurately extract the plane features of the room and corridor from the global point cloud by adopting the method of plane fitting and geometric feature analysis. This method not only greatly reduces the consumption of computing resources, but also ensures the high accuracy of the segmentation result through multiple iterative optimizations, and can accurately restore the geometric structures of the room and corridor. In addition, the method provided in this embodiment still has extremely high robustness when facing noisy point cloud data, can stably complete the segmentation task, and is particularly suitable for real-time point cloud plane feature division in complex environments, providing reliable technical support for related application scenarios.

[0033] In some embodiments, using the structured point cloud for environment segmentation and topology construction to obtain a global topology map and performing global path planning based on the global topology map to obtain a target path, including: performing environment segmentation on the structured point cloud including different room plane point clouds and corridor plane point clouds to obtain nodes representing different regions, and connecting the nodes to represent the connectivity between the regions to obtain a global topology map; performing global path planning on the first node as the starting point and the second node as the final target node among the nodes based on the global topology map to obtain an optimal target path.

[0034] In one embodiment, the structured point cloud including the room plane point clouds of different rooms and the corridor plane point cloud is subjected to environment segmentation to obtain nodes representing different regions, and the nodes are connected to represent the connectivity relationship between the regions to obtain a global topology map, including: regarding the room plane point clouds of different rooms and the corridor plane point cloud as different nodes, that is, all the plane point clouds of one room are regarded as one node, and the corridor plane point cloud corresponding to the outside of one room is regarded as another node, and different nodes are sequentially connected based on the connectivity relationship between the corridor and the room to form different edges, so as to obtain the global topology map. In this embodiment, by abstracting different environments such as rooms and corridors into key nodes and topological connection relationships, the solution scale of path planning can be significantly reduced, thereby ensuring that the intelligent device applying this application has both real-time response ability and global optimality in complex scenarios.

[0035] In one embodiment, global path planning is performed on the first node as the starting point and the second node as the final target node among the nodes based on the global topology map to obtain a target path, including: determining the first node as the starting point and the second node as the target node among the nodes, calculating the shortest path estimated values of the remaining nodes including the second node and the first node as the starting point, and storing the remaining nodes including the second node according to the magnitudes of the shortest path estimated values to obtain a priority queue, that is, arranging them from small to large, and the node corresponding to the smallest shortest path estimated value is located at the head of the priority queue; initializing the shortest path of the first node to zero and the shortest paths of the remaining nodes and the first node to infinity, taking out the element at the head of the priority queue, that is, the third node corresponding to the minimum shortest path estimated value of the first node, and marking the third node as processed; traversing all the adjacent nodes of the third node, and for each adjacent node, calculating the first path length from the first node passing through the third node to its adjacent node; comparing the first path length with the shortest path estimated value of the adjacent node of the first node and the third node, if the first path length is less than the shortest path estimated value, it means that the path from the first node passing through the third node to its adjacent node is better than the path from the first node directly to the adjacent node of the third node, updating the shortest path estimated value of the first node to the adjacent node of the third node to the first path length, and updating its priority in the priority queue; taking the third node as the new starting point, traversing the remaining unprocessed nodes and repeating the above steps until the processing of the second node as the final target node is completed to obtain the shortest path from the target node to the starting point, and then obtaining the shortest path from the starting point to the target node (i.e., the optimal target path) by backtracking.

[0036] It should be noted that in this embodiment, the shortest path node that is currently known is selected each time and the path is expanded based on this. By gradually updating the shortest path lengths from the starting point to other nodes, the global optimal solution is finally obtained. Therefore, when updating the shortest path of a certain node, the predecessor node (that is, from which node the shortest path to this node comes) will be recorded at the same time. The records of these predecessor nodes form a "path chain" pointing from the target node to the starting point, so that the target path from the starting point to the target node can be constructed forward based on the "path chain" by tracing back from the target node in the reverse direction.

[0037] This embodiment provides a global path planning method based on a structured topology graph, which can efficiently calculate the shortest path, reduce redundancy, and reduce the computational complexity, significantly improving the navigation efficiency of the exploration system applying this embodiment in a complex environment, especially performing well in large-scale indoor and dynamic scenarios. In addition, the method provided in this embodiment also has good dynamic adaptability. When the environment changes, the topology graph can be quickly updated to ensure the real-time performance and accuracy of path planning.

[0038] Furthermore, in the traditional global path planning method, the acquisition field of view of the lidar is fixed, resulting in inefficient global path planning and perspective adjustment in some dynamically changing environments. For this reason, in the above global path planning process of this application, it further includes step S230: dynamically adjusting the acquisition field of view of the lidar at the current moment according to a preset adaptive field of view adjustment strategy, and controlling the lidar to acquire the second local point cloud with the adjusted acquisition field of view; step S240: optimizing the target node by using the second local point cloud to obtain a target exploration point, and generating an exploration path based on the target exploration point.

[0039] Specifically, the acquisition field of view range of the lidar is adjusted in real time through a preset field of view adjustment strategy, and then the planned path of the exploration system in the environment where it is located is adjusted in real time. In other words, the exploration system adjusts the acquisition field of view range of the lidar in real time according to the current task requirements and the complexity of the surrounding environment, that is, by expanding the acquisition field of view of the lidar in complex environment areas and narrowing the acquisition field of view of the lidar in simple environment areas, focusing on the current task, thereby improving the exploration efficiency.

[0040] In one embodiment, the preset adaptive field of view adjustment strategy includes: obtaining the minimum acquisition field of view range of the lidar 、the maximum acquisition field of view range 、the adjustment parameter 、the distance between the lidar and the target node at the current moment and the maximum explorability distance of the lidar , and adjusting the acquisition field of view range of the lidar at the current moment through a preset adjustment formula , where the preset adjustment formula is:

[0041] The above formula adapts to the complexity of the environment by controlling the acquisition field of view of the lidar, ensuring that the exploration system can narrow the acquisition field of view to improve efficiency when approaching the target, and expand the field of view in complex areas to ensure better exploration results.

[0042] In addition, this embodiment can also improve the exploration efficiency through a reinforcement learning algorithm. Specifically, that is, the exploration system dynamically adjusts the preset adaptive field of view adjustment strategy in real time according to the environment in which the exploration system is located, so that the dynamic adjustment of the acquisition field of view of the lidar in the exploration system is more flexible and intelligent, thereby improving the exploration efficiency.

[0043] To avoid the instability caused by the frequent switching of the perspective of the exploration system equipped with a lidar in a dynamic environment, this embodiment optimizes the target node by using the second local point cloud to obtain the target exploration point, and generates an exploration path based on the target exploration point, specifically including: analyzing the environmental position, obstacle distribution position and target node position of the exploration system at the current moment by using the second local point cloud, dynamically adjusting the target node according to the environmental position, obstacle distribution position and target node position of the exploration system at the current moment to obtain the exploration target node, and then optimizing the target path generated based on the target node and generating a smooth and collision-free exploration path according to the target exploration point.

[0044] This embodiment dynamically adjusts the range of the local acquisition field of view of the lidar by real-time obtaining the scene segmentation information, enabling the exploration system applying this method to accurately focus on the local area while performing global exploration, thereby improving the efficiency and accuracy of local exploration. Especially in scenarios with frequent perspective switching and dynamic environmental changes, it can quickly adapt and optimize the path planning.

[0045] In summary, this application constructs a structured topology map to accurately extract environmental structure information such as rooms and corridors, and performs global path planning based on the environmental structure information. By optimizing the global path through the local acquisition field of view dynamic adjustment strategy, it can not only effectively reduce the computational complexity, maintain high stability and accuracy in a dynamic environment, but also enable the system to better adapt to large-scale, complex and dynamically changing scenarios, realizing efficient and accurate autonomous exploration.

[0046] Figure 3It is a schematic block diagram of an active exploration device 300 based on a structured environment provided by an embodiment of the present application. The device 300 includes: an acquisition unit 310, configured to acquire multiple frames of first local point cloud data collected by a lidar at historical moments and attitude information recorded by an IMU sensor at corresponding historical moments, and splice the multiple frames of first local point cloud data using the attitude information corresponding to different historical moments to obtain a global point cloud; a division unit 320, configured to perform plane fitting on the global point cloud to obtain point cloud plane features, and perform feature division on the point cloud plane features through preset geometric features to obtain structured point clouds, where the structured point clouds include different room plane point clouds and corridor plane point clouds; a planning unit 330, configured to perform environment segmentation and topology construction using the structured point clouds to obtain a global topology map and perform global path planning based on the global topology map to obtain a target path.

[0047] For a detailed description of the device 300 proposed in the present application, reference may be made to the foregoing embodiments, and the same content will not be repeated here. In some possible embodiments, the processor in the above-mentioned intelligent device 100 may include the above-mentioned device 300.

[0048] Figure 4 It is a schematic block diagram of a computer device 400 provided by an embodiment of the present application. Figure 4 The computer device 400 shown includes a memory 410, a processor 420, and a bus 440. Optionally, the computer device 400 further includes a communication interface 430. Among them, the memory 410, the processor 420, and the communication interface 430 are communicatively connected to each other through the bus 440.

[0049] The memory 410 may be a read only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 410 may store a program, and when the program stored in the memory 410 is executed by the processor 420, the processor 420 is configured to execute each step of the method proposed by the embodiment of the present application.

[0050] The processor 420 may adopt a general-purpose central processing unit (CPU), a microprocessor, an application specific integrated circuit (ASIC), a graphics processing unit (GPU), or one or more integrated circuits, and is configured to execute relevant programs to implement the method proposed by the embodiment of the present application.

[0051] The processor 420 may also be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the method described in the embodiments of the present application may be completed by the integrated logic circuit in the hardware of the processor 420 or by instructions in the form of software.

[0052] The above-mentioned processor 420 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method involved in the embodiments of the present application may be directly embodied as being executed and completed by the hardware decoding processor, or completed by a combination of the hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory 410, and the processor 420 reads the information in the memory 410 and combines its hardware to complete Figure 3 the functions required to be executed by the units included in the device shown, or execute the method described in the method embodiments of the present application.

[0053] The communication interface 430 uses a transceiver device such as, but not limited to, a transceiver to implement the communication between the device 400 and other devices or communication networks.

[0054] The bus 440 may include a path for transmitting information between various components of the device 400 (for example, the memory 410, the processor 420, the communication interface 430).

[0055] It should be understood that although the above device 400 only shows a memory, a processor, and a communication interface, in the specific implementation process, those skilled in the art should understand that the device 400 may also include other devices necessary for normal operation. At the same time, according to specific needs, those skilled in the art should understand that the device 400 may also include hardware devices for implementing other additional functions. In addition, those skilled in the art should understand that the device 400 may also only include the devices necessary for implementing the embodiments of the present application, and do not necessarily include Figure 4 all the devices shown.

[0056] It should be understood that the processor in the embodiments of the present application may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0057] It should also be understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0058] Figure 5 It is a schematic block diagram of a computer-readable storage medium 500 provided by the embodiments of the present application. Figure 5The computer-readable storage medium 500 shown stores computer instructions 510. When the computer instructions 510 are executed by a processor, the methods corresponding to the above embodiments can be implemented.

[0059] In some possible embodiments, the computer-readable storage medium 500 can be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0060] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0061] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0062] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0063] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0064] In addition, in each embodiment of this application, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0065] When the above-mentioned functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application.

[0066] As described above, the above are only specific implementation manners of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, and all should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. An active exploration method based on a structured environment, characterized in that: Applied to an exploration system including a laser radar and an IMU sensor, the method includes: Acquire multiple frames of first local point cloud data collected by the laser radar at historical moments and the posture information recorded by the IMU sensor at corresponding historical moments, and use the posture information corresponding to different historical moments to splice the multiple frames of first local point cloud data to obtain a global point cloud; Performing plane fitting on the global point cloud to obtain point cloud plane features, and performing feature division on the point cloud plane features by using preset geometric features to obtain a structured point cloud, wherein the structured point cloud includes plane point clouds of different rooms and a plane point cloud of a corridor; The structured point cloud is used to perform environment segmentation and topology construction to obtain a global topology map, and global path planning is performed based on the global topology map to obtain a target path.

2. The active exploration method according to claim 1, characterized in that: The step of performing plane fitting on the global point cloud to obtain the point cloud plane features includes: Selecting any three points in the global point cloud as sample points to construct a first plane and fitting a first plane equation of the first plane, calculating the distances from the remaining points in the global point cloud to the first plane based on the first plane equation, and if the distances are less than a preset threshold, considering the corresponding points as inliers to record the number of first inliers and a first inlier set; If the number of first inliers recorded at the current moment is greater than the number of first inliers recorded at the previous moment, the plane equation is updated based on the first inlier set corresponding to the current moment, and the step of "calculating the distances from the remaining points in the global point cloud to the first plane based on the first plane equation, and if the distance is less than a preset threshold, the corresponding points are regarded as inliers to record the number of first inliers and the first inlier set" is repeated until all points in the complete point cloud are traversed to obtain the second inlier set corresponding to the largest number of inliers; Select any three inner points in the second inner point set to construct an initial plane equation of the second plane, determine the plane parameters of the initial plane equation by minimizing the distance from each inner point in the second inner point set to the second plane to obtain the second plane equation, and regard the plane corresponding to the second plane equation as a point cloud plane feature.

3. The active exploration method according to claim 2, characterized in that: The step of performing feature division on the plane features of the point cloud by using preset geometric features to obtain a structured point cloud includes: The preset geometric features that the room has multiple parallel walls and floors and the corridor is long and straight are used to divide the point cloud plane features into point clouds of different attribute types to obtain a structured point cloud including a room plane point cloud and a corridor plane point cloud.

4. The active exploration method according to claim 1, characterized in that: The method of using structured point cloud to perform environment segmentation and topology construction to obtain a global topology map and performing global path planning based on the global topology map to obtain a target path includes: The structured point clouds including the plane point clouds of different rooms and the plane point clouds of corridors are segmented to obtain nodes representing different areas, and the nodes are connected to represent the connectivity relationship between the areas to obtain a global topological map; Based on the global topology map, global path planning is performed on a first node as a starting point and a second node as a final target node among the nodes to obtain the target path.

5. The active exploration method according to claim 4, characterized in that: The structured point cloud including the plane point clouds of different rooms and the plane point clouds of corridors is segmented to obtain nodes representing different areas, and the nodes are connected to represent the connectivity relationship between the areas to obtain a global topology map, including: The room plane point clouds and corridor plane point clouds belonging to different rooms are regarded as different nodes. All plane point clouds of a room are regarded as one node, and the plane point cloud of a corridor corresponding to a room is regarded as another node. Based on the connectivity relationship between the corridors and the rooms, different nodes are connected in sequence to form different edges, so as to obtain the global topology graph.

6. The active exploration method according to claim 4, characterized in that: The method of performing global path planning on a first node as a starting point and a second node as a final target node in each node based on the global topology map to obtain a target path includes: Determine a first node as a starting point and a second node as a target node among the nodes, and calculate the shortest path estimation value between the remaining nodes including the second node and the first node as the starting point; The remaining nodes including the second node are stored according to the shortest path estimation value to obtain a priority queue, and the node corresponding to the smallest shortest path estimation value is placed at the head of the priority queue; Initialize the shortest path of the first node to zero and the shortest paths of the remaining nodes and the first node to infinity, take out the third node as the head element from the priority queue, and mark the third node as processed; Traversing all adjacent points of the third node, and for each adjacent point, calculating a first path length from the first node to its adjacent node through the third node; Compare the first path length with the shortest path estimation value of the first node and the adjacent point of the third node, and if the first path length is less than the shortest path estimation value, update the shortest path estimation value from the first node to the adjacent point of the third node to the first path length, and update the priority in the priority queue; Taking the third node as a new starting point, traverse the remaining unprocessed nodes until the second node as the target node is processed to obtain the shortest path from the target node to the starting point, and obtain the target path from the starting point to the target node by tracing back.

7. The active exploration method according to any one of claims 1 to 6, characterized in that: The method further comprises: Dynamically adjust the laser radar's current acquisition field of view according to a preset adaptive field of view adjustment strategy, and control the laser radar to acquire a second local point cloud with the adjusted acquisition field of view; The target node is optimized using the second local point cloud to obtain a target exploration point, and an exploration path is generated based on the target exploration point.

8. The active exploration method according to claim 7, characterized in that: The adaptive field of view adjustment strategy includes: Get the minimum acquisition field of view of the laser radar , Maximum acquisition field of view , Adjustment parameters , the distance between the laser radar and the target node at the current moment And the maximum explorable distance of the laser radar , adjust the laser radar's current acquisition field of view through a preset adjustment formula , wherein the preset adjustment formula is:

9. The active exploration method according to claim 7, characterized in that: The step of optimizing the target node by using the second local point cloud to obtain a target exploration point and generating an exploration path based on the target exploration point includes: The second local point cloud is used to analyze the current environmental position, obstacle distribution position and target node position of the exploration system, and the target node is dynamically adjusted according to the current environmental position, obstacle distribution position and target node position of the exploration system to obtain the exploration target node, and a smooth and collision-free exploration path is generated according to the exploration target point.

10. An active exploration device based on a structured environment, characterized in that: include: An acquisition unit is used to acquire multiple frames of first local point cloud data collected by the laser radar at historical moments and the posture information recorded by the IMU sensor at corresponding historical moments, and to splice multiple frames of first local point cloud data using the posture information corresponding to different historical moments to obtain a global point cloud; A division unit, used for performing plane fitting on the global point cloud to obtain point cloud plane features, and performing feature division on the point cloud plane features by preset geometric features to obtain a structured point cloud, wherein the structured point cloud includes plane point clouds of different rooms and a plane point cloud of a corridor; A planning unit is used to use the structured point cloud to perform environment segmentation and topology construction to obtain a global topology map and perform global path planning based on the global topology map to obtain a target path.

11. An active exploration system based on a structured environment, characterized in that: The active exploration system comprises: LiDAR is used to scan the indoor environment including corridors and multiple rooms where the exploration system is located at different times to obtain local point cloud data from different perspectives; An IMU sensor is used to record the attitude information of the exploration system while the laser radar collects point cloud data; A processing platform is used to process the local point cloud data and the posture information at different times according to the active exploration method according to any one of claims 1 to 9 to obtain a target path.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the active exploration method according to any one of claims 1 to 9.

Images