SLAM-based mapping navigation method and mobile measurement backpack

Abstract

The invention provides an SLAM (Simultaneous Localization and Mapping)-based mapping navigation method and a mobile measurement backpack, and the method comprises the following steps: based on an SLAM algorithm, fusing laser point cloud information and image texture information, and constructing a three-dimensional environment map of a current position; environment map view data of the current position are obtained, a local three-dimensional map containing navigation obstacle avoidance information is constructed in real time based on the three-dimensional environment map, and the area where the local three-dimensional map is located is a preset range area in the area where the three-dimensional environment map is located; identifying an unknown area which is not explored in the local three-dimensional map, and determining an exploration priority according to the importance of the unknown area; and generating a navigation path by using a path planning algorithm according to the exploration priority. The problem that in the prior art, in the walking data collection process of an SLAM backpack, data are prone to being repeatedly collected, and the working efficiency is reduced is solved.

Description

Technical Field

[0001] The present application relates to the technical field of digital map navigation, and in particular to a SLAM-based mapping navigation method and a mobile measurement backpack. Background Art

[0002] In early robot navigation systems, positioning and mapping were usually performed separately. First, the environment map was constructed using sensor data, and then positioning was performed based on the map. This method has some problems, such as the need to provide detailed map information in advance, and there is feedback coupling between positioning and mapping, which affects system performance. With the hardware upgrade of sensors and the continuous optimization of position estimation and map representation algorithms, SLAM algorithms (Simultaneous Localization and Mapping) have gradually become the first choice for current mobile navigation systems. SLAM refers to the process in which a positioning device such as a robot or mobile device simultaneously locates its own position and constructs an environmental map in an unknown environment by combining sensor data. It transforms the traditional problem of separating positioning and mapping into a mutually dependent relationship between the two. Among them, solving the positioning information requires information about the current environment, and the quality of mapping also depends on the accuracy of the positioning information. By continuously optimizing the robot's position estimation and environmental map, two-way improvements in positioning and mapping are ultimately achieved.

[0003] Many existing devices use SLAM algorithms for positioning and mapping. For example, SLAM backpacks are mainly used in scenarios such as 3D reconstruction, mapping, and navigation in indoor and outdoor environments. Such as digital work in the fields of architecture, infrastructure, and cultural relics protection. The current performance of many SLAM backpacks is not very good. Existing SLAM backpacks usually rely on people to carry and walk to collect data, and the data collection itinerary is determined by human will, which easily leads to repeated data collection, which not only reduces work efficiency and increases data collection costs, but is also not intelligent enough.

[0004] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the purpose of this application is to provide a SLAM-based mapping and navigation method and a mobile measurement backpack, which solves the problem in the prior art that SLAM backpacks usually easily cause repeated data collection and reduce work efficiency during the process of walking and collecting data.

[0006] On the one hand, the present application provides a method for mapping and navigation based on SLAM, comprising the steps of:

[0007] Based on the SLAM algorithm, the laser point cloud information and image texture information are integrated to construct a three-dimensional environment map of the current location;

[0008] Obtaining the environmental map data of the current position, and constructing a local three-dimensional map containing navigation and obstacle avoidance information in real time based on the three-dimensional environmental map, wherein the area where the local three-dimensional map is located is a predetermined range area within the area where the three-dimensional environmental map is located;

[0009] Identify unknown areas that have not been explored in the local 3D map and determine the exploration priority based on the importance of the unknown areas;

[0010] Based on the exploration priority, a navigation path is generated using a path planning algorithm.

[0011] Optionally, based on the SLAM algorithm, the laser point cloud information and the image texture information are integrated to construct a three-dimensional environment map, including:

[0012] Obtain laser point cloud information through laser radar and image texture information through visual sensor;

[0013] Extracting feature points from the acquired laser point cloud information and the acquired image texture information, and establishing a corresponding relationship between the feature points of the laser point cloud information and the feature points of the image texture information;

[0014] According to the correspondence between the feature points, the relative position and posture of the lidar and the relative position and posture of the visual sensor are estimated;

[0015] Based on the relative position and posture of the laser radar and the relative position and posture of the visual sensor, the laser point cloud information and the image texture information are fused to construct a three-dimensional environment map.

[0016] Optionally, before the step of acquiring laser point cloud information through a laser radar and acquiring image texture information through a visual sensor, the step further includes:

[0017] Reference calibration is performed between multiple laser radars and between each laser radar, visual sensor and inertial measurement unit to determine the geometric relationship between multiple laser radars, visual sensors and inertial measurement units.

[0018] Optionally, after fusing the laser point cloud information and the image texture information according to the relative posture to construct a three-dimensional environment map, the step further includes:

[0019] Continuously obtain the laser point cloud information and image texture information of the next position, and build and update the relative position and three-dimensional environment map of the next position.

[0020] Optionally, the step of obtaining the environmental map data of the current position and constructing a local three-dimensional map containing navigation and obstacle avoidance information in real time based on the three-dimensional environmental map includes:

[0021] Obtain the current area's environmental image data through the depth camera;

[0022] According to the image depth information in the environmental map data, an obstacle information map containing navigation and obstacle avoidance information of the current area is constructed.

[0023] The obstacle information map is integrated with the three-dimensional environment map to obtain a local three-dimensional map.

[0024] Optionally, in the step of constructing an obstacle information map containing navigation and obstacle avoidance information of the current area in real time through environmental map data based on a visual SLAM algorithm:

[0025] Navigation obstacle avoidance information includes environmental objects, obstacles, and passable area information.

[0026] Optionally, in the step of generating a navigation path using a path planning algorithm according to the exploration priority:

[0027] The path planning algorithm adopts Dijkstra path planning algorithm.

[0028] Optionally, the step of generating a navigation path using a path planning algorithm according to the exploration priority further includes the following steps:

[0029] Display the generated navigation path.

[0030] On the other hand, the present application also proposes a mobile measurement backpack, wherein the SLAM-based mapping and navigation method as described above is applied, wherein the mobile measurement backpack includes:

[0031] A support frame, to which a backpack body is connected, and the backpack body is used to be placed on the back of the operator;

[0032] Laser radar and visual sensor, the laser radar and visual sensor are arranged on the supporting frame;

[0033] The display controller is connected to the side of the support frame away from the backpack body and is used to be set in front of the operator's head.

[0034] Optionally, the laser radar and the visual sensor are combined into a measurement data acquisition module, and the measurement data acquisition module is detachably connected to the support frame;

[0035] The display controller is movably connected to the support frame through the front frame, and the display controller can be flipped in the up and down directions by bending.

[0036] Beneficial effects: A method for mapping and navigation based on SLAM and a mobile measurement backpack in the present application can well express the structural information of the three-dimensional space according to the acquired laser point cloud information, but it is weak in texture and color information, and the acquired image texture information has good visual expression ability, but its depth perception ability is weak. The three-dimensional environment map obtained by fusing laser point cloud information and image texture information has the advantages of good depth perception effect and good visual expression effect, so the obtained three-dimensional environment map can obtain more complete environmental perception information and improve the accuracy and robustness of SLAM. On the basis of establishing a three-dimensional environment map, the local three-dimensional map constructed after obtaining the environmental map data of the current position makes the map have more accurate navigation and obstacle avoidance information, which can be used for unknown space exploration and autonomous path planning, making the planned navigation path more reasonable and efficient. Through the calculation and verification of the algorithm, the integrity of the three-dimensional environment mapping is improved, and the work efficiency is comprehensively improved. According to the exploration priority planning on the local three-dimensional map with more accurate navigation and obstacle avoidance information, the intelligent generation of intelligent navigation paths in unknown areas that have not been explored can be realized, which can further avoid the problems of repeated data collection and incomplete three-dimensional space collection caused by artificial will planning paths. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flowchart of the main steps of a SLAM-based mapping and navigation method according to an embodiment of the present application;

[0038] Figure 2 A flowchart of the detailed steps of a SLAM-based mapping and navigation method according to an embodiment of the present application;

[0039] Figure 3 This is a schematic diagram of the structure of a mobile measurement backpack according to an embodiment of the present application;

[0040] Figure 4 This is a schematic diagram of the structure of a measurement data acquisition module of a mobile measurement backpack according to an embodiment of the present application;

[0041] Figure 5 A schematic structural diagram of a mobile measurement backpack from another perspective of an embodiment of the present application;

[0042] Figure 6 A right side view of a mobile measurement backpack according to an embodiment of the present application;

[0043] Figure 7 This is a schematic structural diagram of a mobile measurement backpack in an embodiment of the present application in a folded-down state;

[0044] Figure 8 This is a reference schematic diagram of a measurement data acquisition module of a mobile measurement backpack in an embodiment of the present application applied in different scenarios.

[0045] The numbers in the figure are: 100, support frame; 110, vertical frame; 111, shoulder bracket; 112, reinforcing rib rod; 120, front frame; 121, rotation limit structure; 122, tilt frame; 123, rotating connecting arm; 130, lighting; 200, measurement data acquisition module; 210, shell; 220, laser radar; 230, movable gimbal; 240, visual sensor; 250, depth camera; 300, display controller; 310, touch screen; 400, backpack body; 410, backpack box; 420, shoulder strap; 430, waist strap; 500, RTK positioning module. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present application clearer and more specific, the present application is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific examples described here are only used to explain the present application and are not used to limit the present application.

[0047] The existing SLAM backpacks not only easily lead to repeated data collection, reduce work efficiency, increase data collection costs, and are not intelligent enough, but the sensor equipment of the existing SLAM backpacks is generally a monocular camera or even does not carry a camera, and only uses low-quality equipment such as single-line laser radar. The single detection hardware leads to reduced performance of the entire system, and the single device cannot provide more accurate positioning information and better mapping effects for confined spaces. In addition, the current SLAM backpacks are all integrated devices, that is, the entire backpack body and the data collection module used for data collection are combined together and cannot be disassembled, which is not conducive to replacing the mobile platform at any time, and cannot solve the problem of data collection by manual movement, and cannot be applied to changing scenarios, such as dangerous areas where no one completes data collection through human movement. In order to optimize the problems of existing SLAM backpacks, this application proposes the following embodiments:

[0048] Embodiment 1

[0049] like Figure 1 As shown, this embodiment proposes a method for mapping and navigation based on SLAM, which can be applied to a mobile measurement backpack (SLAM backpack), where an operator (operator) carries the mobile measurement backpack on his back and measures environmental data through manual movement, thereby realizing functions such as positioning, mapping, and navigation. Or it can be applied to mobile intelligent mechanical equipment (such as robots and robot dogs), where environmental data can be measured through mechanical automatic movement.

[0050] like Figure 1 As shown, the present application provides a method for mapping and navigation based on SLAM, comprising the following steps:

[0051] Step S100: Based on the SLAM algorithm, the laser point cloud information and the image texture information are integrated to construct a three-dimensional environment map of the current location.

[0052] In the specific process, based on the characteristics of laser radar that is good at measuring the three-dimensional distance information of the environment, laser radar is used to obtain laser point cloud information. The laser point cloud information obtained can well express the structural information of the three-dimensional space, but the laser radar is weak in texture and color information. The visual sensor is good at obtaining high-resolution color and texture information of the environment. Therefore, the image texture information obtained by the visual sensor, such as RGB images, has good visual expression ability, but its depth perception ability is weak. By integrating the advantages of the two sensors, more complete environmental perception information can be obtained, improving the accuracy and robustness of SLAM.

[0053] In the detection system of the SLAM algorithm, one or more laser radars can be set. In order to improve the accuracy of the laser point cloud information, multiple laser radars are set in this embodiment. The multiple laser radars work together to detect in different directions and obtain laser point cloud information in different directions, which expands the mapping range and improves the regional range of the three-dimensional environment map.

[0054] like Figure 1 , Figure 2 As shown, step S100 in this embodiment specifically includes the following steps:

[0055] Step S110: perform reference calibration between the multiple laser radars and between each laser radar, visual sensor and inertial measurement unit to determine the geometric relationship between the multiple laser radars, visual sensors and inertial measurement units.

[0056] In the specific process, it is necessary to calibrate the internal and external parameters between multiple laser radars, and between each laser radar and visual sensor, and determine the geometric relationship between each sensor (all laser radars and visual sensors). In this way, the calibration of multiple laser radars and visual sensors can be realized, so that multiple laser radars and visual sensors can measure data in a unified system coordinate system, improve the matching of position data, facilitate the subsequent unified processing of position data, and improve accuracy.

[0057] Step S120: acquiring laser point cloud information through a laser radar, and acquiring image texture information through a visual sensor.

[0058] In the specific process, the visual sensor uses a three-axis gimbal camera, and the camera can be an infrared camera, which improves the image clarity and makes the image texture information more accurate. Therefore, this embodiment uses three laser radars and one three-axis gimbal infrared camera for data measurement. Among them, the three laser radars are located on the left, right and back sides, 90 degrees to each other, and the laser radars are placed vertically, which increases the sensing area, increases the sensing area, and the three laser radars work together in three different directions to improve data accuracy. The top is a three-axis gimbal camera, which uses an industrial camera with high shooting accuracy and fast response speed. It can detect and obtain more accurate image texture information during the operator's movement. The camera can be controlled to take pictures through the three-axis gimbal, and pictures can be taken while the operator is moving, thereby obtaining image texture information. By fusing information from different sensors, the accuracy and robustness of positioning and map construction are improved.

[0059] After obtaining the above data, the laser point cloud information obtained by the lidar, the image texture information obtained by the visual sensor, and the IMU data obtained by the inertial measurement unit are synchronously calibrated, so that the frames of image texture information, IMU data, and laser point cloud information received at the same time are one-to-one corresponding.

[0060] Step S130: extract feature points from the acquired laser point cloud information and the acquired image texture information, and establish a corresponding relationship between the feature points of the laser point cloud information and the feature points of the image texture information.

[0061] Step S140: Estimate the relative position and posture of the laser radar and the visual sensor based on the correspondence between the feature points.

[0062] In the specific process, the correlation between the laser point cloud information and the acquired image texture information is determined by extracting feature points. Based on the correspondence between the feature points, the relative position and posture of the laser radar and the visual sensor are estimated using an optimization algorithm, and the relative position and posture can be used for positioning. In order to make the calculation of the relative position and posture more accurate, it is usually combined with an inertial measurement unit (IMU) for detection.

[0063] In the specific process, the specific process of the above steps S130-S140 is as follows:

[0064] The image texture information obtained by the visual sensor and the IMU data obtained by the inertial measurement unit are processed by VIO (visual inertial odometry), and the laser point cloud information obtained by the lidar and the IMU data obtained by the inertial measurement unit are processed by LIO (lidar inertial odometry).

[0065] In the LIO (LiDAR Inertial Odometry) processing: Use the back propagation method to remove motion distortion in the laser point cloud information, extract feature points based on point cloud curvature segmentation, and construct the point cloud residual equation. Use the distance from the dedistorted point cloud to the plane for frame-to-map matching. Forward propagate the IMU data and construct the IMU residual equation.

[0066] In the VIO (Visual Inertial Odometry) processing: based on the current field of view, the currently observed visual sub-map is selected from the global visual map, and the occluded and depth discontinuous points are removed, and then the image photometric error from the frame to the map is calculated based on the sparse optical flow method. Finally, the residual from the point to the plane and the visual photometric error and the forward propagation value of the IMU are put into the error state iterative Kalman filter to obtain the accurate pose, and the pose is used to update the newly observed data to the map. Among them, the visual global map is a collection of previously observed laser point clouds.

[0067] Step S150: Based on the relative position of the laser radar and the relative position of the visual sensor, the laser point cloud information and the image texture information are integrated to construct a three-dimensional environment map. In the specific process, based on the optimized relative position of each sensor, the point cloud data of the current frame is added to the visual global map.

[0068] Step S160: continuously obtain the laser point cloud information and image texture information of the next position, and construct and update the relative position and three-dimensional environment map of the next position.

[0069] In the specific process, as new data is continuously acquired during movement, the relative posture and three-dimensional environment map are continuously optimized using an optimized SLAM algorithm to optimize and update the map in real time.

[0070] Step S200, obtain the environmental map data of the current position, and construct a local three-dimensional map containing navigation and obstacle avoidance information in real time based on the three-dimensional environmental map. The area where the local three-dimensional map is located is a predetermined range area in the area where the three-dimensional environmental map is located.

[0071] During the specific process, this embodiment also obtains environmental map data through a depth camera. The detection area set by the depth camera is located in the front. The depth camera can obtain a more accurate and detailed map of a smaller area in a larger three-dimensional environmental map. In this way, in a small area at the current location, more accurate unknown space exploration and autonomous path planning can be carried out based on more comprehensive navigation and obstacle avoidance information.

[0072] like Figure 1 , Figure 2 As shown, step S200 of this embodiment specifically includes the following steps:

[0073] Step S210: Acquire environmental image data of the current area through a depth camera.

[0074] In the specific process, a depth camera (such as a visual sensor such as a camera) is used to obtain image or video data of the current environment. In addition, the image or video data of the current environment can be processed to obtain environmental image data, which contains image depth information. For example, deep learning technology is used to improve the performance of feature extraction and data association, so that the image or video data forms more standard environmental image data with more obvious features and better association.

[0075] Step S220: construct an obstacle information map containing navigation and obstacle avoidance information for the current area according to the image depth information in the environmental map data.

[0076] In the specific process, the environmental depth information collected by the depth camera is processed to obtain the corresponding obstacle information map, which contains navigation and obstacle avoidance information. The navigation and obstacle avoidance information contains the main information of the environmental contour, such as environmental objects, obstacles, passable area information, etc. Through this precise information, the system knows how to avoid obstacles, thereby providing a reference for path planning.

[0077] Step S230: merge the obstacle information map with the three-dimensional environment map to obtain a local three-dimensional map.

[0078] In the specific process, the obstacle information map is further integrated with the three-dimensional environment map, and the obtained local three-dimensional map contains information such as environmental objects, obstacles, and passable areas. Combining the advantages of both, not only does the local three-dimensional map have more accurate environmental contour features, but also the overall path planning can be performed in the three-dimensional environment map. For example, after the path planning is performed in a larger range of the three-dimensional environment map, the path is fine-tuned according to the small-scale local three-dimensional map, which can not only obtain more complete environmental perception information and improve the efficiency of route planning, but also reduce the computational burden and improve the response speed.

[0079] Step S300: Identify unknown areas that have not been explored in the local three-dimensional map, and determine the exploration priority according to the importance of the unknown areas.

[0080] In the specific process, the importance of the unknown area can be determined based on factors such as distance and explorability. The factors that affect the importance of the unknown area are numerically quantified, and different factors are assigned corresponding weights. The exploration priority is determined based on the final sum value. The larger the sum value, the higher the exploration priority.

[0081] Step S400: Generate a navigation path using a path planning algorithm according to the exploration priority.

[0082] In the specific process, the path planning algorithm uses the Dijkstra path planning algorithm to generate a navigation path.

[0083] Step S500: display the generated navigation path.

[0084] During the specific estimation process, the generated navigation path can be pushed to the display screen in front for display, so that the operator can see the map information, positioning information and planned path information, thereby guiding the operator to move according to the path.

[0085] The display screen is not only used to display the 3D environment map rendered by matching the image and point cloud, but also can control the posture of the three-axis gimbal camera above by touching the screen, so as to achieve refined mapping of the target area. In addition, through the display screen, the operator can freely manually control the camera shooting, and the generated RGB image can be saved and used for later detection tasks.

[0086] If a smart mobile device such as a robot is used, there is no need to use a display screen for control.

[0087] Step S600: continuously update the local three-dimensional map by moving the position, and make the next navigation decision according to the current local three-dimensional map and the navigation target.

[0088] Through the above steps, it is possible to autonomously build maps, plan paths and navigate in unknown environments. Through continuous perception, decision-making and execution, the mobile platform can continuously explore the space where data needs to be collected. Through the calculation and verification of the algorithm, the integrity of the three-dimensional environment map can be ensured, and the work efficiency can be comprehensively improved. In this way, the problem of repeated data collection and incomplete three-dimensional space collection caused by artificial will planning of paths can be further avoided.

[0089] In addition, in this embodiment, positioning is also performed through an RTK positioning module, which can achieve accurate positioning without a GNSS signal.

[0090] Embodiment 2

[0091] like Figure 3 , Figure 4As shown, the present application also proposes a mobile measurement backpack, in which the SLAM-based mapping and navigation method described above is applied. The mobile measurement backpack mainly includes: a support frame 100, a laser radar 220 and a visual sensor 240. The support frame 100 is connected to a backpack body 400, and the backpack body 400 is used to be set on the back of the operator. The laser radar 220 and the visual sensor 240 are arranged on the support frame 100, and the laser point cloud information is obtained by the laser radar 220. The visual sensor 240 is generally a camera, and the image texture information is obtained by the visual sensor 240. The display controller 300 is connected to the side of the support frame 100 away from the backpack body 400, and is used to be set in front of the operator's head.

[0092] The laser radar 220 and the visual sensor 240 in this embodiment are combined into a measurement data acquisition module 200. The measurement data acquisition module 200 can be directly fixed on the top of the support frame 100, or can be detachably set on the top of the support frame 100. The measurement data acquisition module 200 in this embodiment can be detachably assembled with the support frame 100 through a disassembly structure. Figure 4 , Figure 8 As shown, according to the needs of the task environment, the measurement data acquisition module 200 can be mounted on a movable carrier such as a human body (handheld or carried on the back), a robot dog, a mobile car, etc., which improves the flexibility of the data acquisition module and the scenario applicability of the overall solution. In particular, when mapping dangerous areas or small spaces, the measurement data acquisition module 200 can be installed on a robot dog or a mobile car, so that the robot dog and the mobile car can enter the dangerous area or small space for mapping, which improves the practicality of the measurement data acquisition module 200.

[0093] The measurement data acquisition module 200 in this embodiment specifically includes: a housing 210, a laser radar 220, a movable pan-tilt platform 230, and a camera. The housing 210 is detachably arranged on the top of the support frame 100, and the housing 210 can be connected to the support frame 100 through a screw connection. For example, a threaded hole is provided on the lower surface of the housing 210, and a screw is provided on the frame, and the screw passes through the frame and is screwed into the threaded hole, thereby realizing a detachable connection between the housing 210 and the support frame 100. In addition to the above-mentioned screw connection, other commonly used detachable connection structures can also be used, such as hook docking, plug-in, etc.

[0094] like Figure 3 , Figure 4As shown, the laser radar 220 is arranged on the housing 210, the movable gimbal 230 is arranged on the housing 210, and the visual sensor 240 (camera) is arranged on the movable gimbal 230. The laser radar 220 and the visual sensor 240 can be integrated in the housing 210 as different sensors. The housing 210 and the support frame 100 are directly detachably installed, which makes the structure simple and the installation more convenient. The laser radar 220 is used to obtain the surrounding laser point cloud information, which can be used for subsequent mapping. The movable gimbal 230 drives the visual sensor 240 to adjust the posture through the control signal of the display manipulator 300. For example, the display manipulator 300 can be operated to control the movement of the movable gimbal 230 above to drive the visual sensor 240 to adjust the posture, thereby realizing the refined mapping of the target area. The operator can also freely manually control the shooting of the visual sensor 240 through the display manipulator 300, and the RGB image generated by it can be saved in the control device and used for later detection tasks.

[0095] like Figure 3 , Figure 4 As shown, further, three laser radars 220 are provided, and the three laser radars 220 are arranged on different sides of the shell 210. In the specific structure, the three laser radars 220 are located on the left side, the right side and the rear side, and are 90 degrees apart from each other, and the laser radars 220 are placed vertically, which increases the sensing area, enlarges the sensing area, and the three laser radars 220 work together in three different directions to improve data accuracy.

[0096] like Figure 3 , Figure 4 As shown, further, the movable gimbal 230 is a three-axis gimbal (the three-axis gimbal is an existing structure and no detailed structural description is given). The three-axis gimbal can drive the camera to achieve multi-angle position movement, thereby obtaining images at different positions, so that relevant image texture information of a larger area can be obtained. The visual sensor 240 of this embodiment is an infrared camera. The infrared camera adopts an industrial camera, which has high shooting accuracy and fast response speed. It performs mapping while the operator is moving to obtain more accurate image texture information. The camera can be controlled to take pictures through the three-axis gimbal, and pictures can be taken while the operator is moving, so as to obtain image texture information, and the surrounding laser point cloud information is obtained by combining the laser radar 220, and the SALM algorithm is used for fusion to achieve positioning and map construction. Therefore, by fusing information from different sensors, the accuracy and robustness of positioning and map construction are improved.

[0097] like Figure 3 , Figure 4As shown, further, a depth camera 250 is provided on the front side of the housing 210 in this embodiment, and the depth camera 250 can use the visual SALM algorithm alone to build a detailed map of a small area in front, which not only makes the map construction more accurate, but also provides a reference for subsequent automatic navigation and obstacle avoidance. The function of the measurement data acquisition module 200 is further improved and more practical.

[0098] like Figure 5 , Figure 6 , Figure 7 As shown, further, the display controller 300 in this embodiment is movably connected to the support frame 100 through the front frame 120, and the display controller 300 is flipped in the up and down direction by bending. The front end of the front frame 120 is connected to the display controller 300, and the rear end is rotatably connected to the support frame 100, so that it can be lifted and lowered around the rotation axis. Thereby, the display operator is lifted and lowered. In order to minimize the interference caused by the display controller 300 in the front to the camera on the three-axis gimbal capturing the image information below, this embodiment solves this problem by designing the display controller 300 to be foldable up and down. Specifically: when the operator controls the three-axis gimbal to shoot downward, the display controller 300 can be adjusted to fold down, so that the display controller 300 will not block the lens of the camera shooting downward. When the camera on the three-axis gimbal does not need to shoot downward, the display screen can be turned upward. During the up and down adjustment and folding, the display controller 300 can work normally.

[0099] like Figure 6 , Figure 7 As shown, further, the pre-frame 120 and the support frame body 100 in this embodiment can be folded by a rotation limiting structure 121, and the rotation limiting structure 121 can adopt an existing conventional structure, for example, the rotation limiting structure 121 can include: a turntable fixed on the pre-frame 120 and arranged coaxially with the rotation axis, and an elastic clamping member arranged on the support frame body 100. In the specific structure, a plurality of clamping holes arranged in a circle are opened at the side of the turntable. When the turntable rotates with the pre-frame 120, the elastic clamping member can be stuck in different clamping holes, so that the pre-frame 120 is limited at different rotation angles to achieve folding up and down. The elastic clamping member needs the turntable to apply a certain extrusion force to get out of the clamping hole, so when the pre-frame 120 is forced to bend, the activity and limitation of the pre-frame 120 can be achieved. In addition to the above-mentioned rotation limiting structure 121, other commonly used rotation limiting structures 121 can also be adopted, such as the rotation structure in the seat handle, etc.

[0100] like Figure 3 , Figure 5As shown, further, the display controller 300 in this embodiment includes a touch screen 310, and the touch screen 310 is fixedly arranged on the front frame 120. The touch screen 310 realizes both the display function and the function of sending control instructions by touch, and adopts human-computer interaction to improve the intelligence of the device. The touch screen 310 is used to display the environment map after the image and point cloud matching rendering. In addition, the posture of the PTZ camera above is controlled by touching the screen, so as to realize the refined mapping of the target area. In addition, the interaction through the touch screen 310 allows the operator to freely manually control the shooting of the camera, and the RGB image generated can be saved in the computer and used for later detection tasks.

[0101] like Figure 5 , Figure 6 As shown, further, the support frame 100 in this embodiment specifically includes: a vertical frame 110 and two shoulder brackets 111. The backpack body 400 is arranged on the back of the vertical frame 110, the measurement data acquisition module 200 is arranged on the top of the vertical frame 110, and the two shoulder brackets 111 are respectively arranged on the left and right sides of the vertical frame 110, and a wearing space for accommodating the operator's head is formed between the shoulder brackets 111 on the left and right sides. The two shoulder brackets 111 can extend to the front of the operator, thereby providing stable support for the relevant equipment in front. Reinforcement rods 112 are respectively arranged between the two shoulder brackets 111 and the vertical frame 110, which improves the support strength of the entire support frame 100 and facilitates the wearing of the support frame 100. The entire support frame 100 can be made of relatively light materials to reduce the walking burden of the operator.

[0102] like Figure 5 , Figure 6 As shown, further, the front frame 120 in this embodiment includes: a tilt frame body 122 and a rotating connecting arm 123. The display controller 300 is set on the tilt frame body 122, and the rotating connecting arm 123 is respectively set at the left and right ends of the tilt frame body 122 and movably connected to the shoulder bracket 111. On the vertical plane in the left and right directions, the tilt frame body 122 is tilted to the rotating connecting arm 123. Since the tilt frame body 122 is tilted, the display controller 300 connected thereto is also tilted, so that it is convenient for people to watch the screen and it is also convenient for operation.

[0103] like Figure 3 , Figure 5As shown, further, a lighting lamp 130 is provided on the support frame 100 in this embodiment. In order to deal with the darker areas in the environment, which affect the image quality captured by the depth camera 250 and the camera in front of the backpack, this embodiment is also equipped with lighting lamps 130 in different directions on the support frame 100. In the specific structure, the lighting lamps 130 can be set on the left and right sides of the support frame 100, for example, the lighting lamp 130 is installed in the triangular area surrounded by the reinforcing rib rod 112, the vertical frame 110 and the shoulder bracket 111. The lighting lamp 130 is installed on the crossbeam (front or / and rear) of the vertical frame 110, for example, below the measurement data acquisition module 200. And the lighting lamp 130 is installed on the crossbeam of the inclined frame 122 below the display controller 300.

[0104] like Figure 3 , Figure 5 , Figure 6 As shown, further, the backpack body 400 of this embodiment specifically includes: a backpack box 410, a shoulder strap 420 and a waist strap 430. The backpack box 410 is connected to the back of the support frame 100, and a storage space is provided in the backpack box 410, and the storage space is used to set the control device and the power supply device. The shoulder strap 420 is connected to the backpack box 410 and is used to be put on the operator's shoulder. The waist strap 430 is connected to the backpack box 410 and is used to be tied to the operator's waist.

[0105] like Figure 3 As shown, the mobile measurement backpack of this embodiment further includes an RTK positioning module 500, which is fixedly arranged on the backpack body 400, and can be specifically arranged vertically in the middle position of the backpack box 410 in the left and right directions, protruding from the backpack box 410. The RTK positioning module 500 adopts mature positioning technology and can achieve accurate positioning without GNSS signals.

[0106] In summary, due to the limitations of existing SLAM backpacks, the present application provides a method for mapping navigation based on SLAM and a close-range mobile measurement backpack with intelligent guidance. By combining a depth camera, an industrial-grade infrared pan-tilt camera and multiple lidar sensors, an RTK positioning module, and an inertial measurement unit, a multi-sensor fusion technical solution is adopted to achieve the effect of precise positioning and accurate mapping. In addition, in order to realize intelligent mobile measurement, on the basis of SLAM technology, intelligent guidance is proposed, the mobile path is determined by intelligent algorithm calculation, and the mobile platform is guided to move forward according to the planned path, so as to realize intelligent guidance of mobile data collection. And through modular design, a variety of different sensors are integrated into a detachable module. According to the needs of the task environment, the module can be carried on movable carriers such as the human body (handheld or carried on the back), a robot dog, and a mobile car, which improves the flexibility of the data acquisition module and the scene applicability of the overall solution.

[0107] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these 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 application.

Claims

1. A method for mapping and navigation based on SLAM, characterized in that: The method comprises the steps of: Based on the SLAM algorithm, the laser point cloud information and image texture information are integrated to construct a three-dimensional environment map of the current location; Acquire environmental map data of the current position, and construct a local three-dimensional map containing navigation and obstacle avoidance information in real time based on the three-dimensional environmental map, wherein the area where the local three-dimensional map is located is a predetermined range area within the area where the three-dimensional environmental map is located; Identifying unknown areas that have not been explored in the local three-dimensional map, and determining exploration priorities according to the importance of the unknown areas; A navigation path is generated using a path planning algorithm according to the exploration priority.

2. The method for mapping and navigation based on SLAM according to claim 1, characterized in that: Based on the SLAM algorithm, the laser point cloud information and image texture information are integrated to construct a 3D environment map. The steps include: Obtain laser point cloud information through laser radar and image texture information through visual sensor; Extracting feature points from the acquired laser point cloud information and the acquired image texture information, and establishing a corresponding relationship between the feature points of the laser point cloud information and the feature points of the image texture information; According to the correspondence between the feature points, the relative position and posture of the laser radar and the relative position and posture of the visual sensor are estimated; Based on the relative position and posture of the laser radar and the relative position and posture of the visual sensor, the laser point cloud information and the image texture information are fused to construct a three-dimensional environment map.

3. The method for mapping and navigation based on SLAM according to claim 1, characterized in that: The steps of obtaining laser point cloud information through laser radar and image texture information through visual sensor also include the following steps: Reference calibration is performed between multiple laser radars and between each laser radar, visual sensor and inertial measurement unit to determine the geometric relationship between multiple laser radars, visual sensors and inertial measurement units.

4. The method for mapping and navigation based on SLAM according to claim 2, characterized in that: After the step of fusing the laser point cloud information and the image texture information according to the relative position and posture to construct a three-dimensional environment map, the following steps are further included: Continuously obtain the laser point cloud information and image texture information of the next position, and build and update the relative position and three-dimensional environment map of the next position.

5. The method for mapping and navigation based on SLAM according to claim 1, characterized in that: The steps of obtaining the environmental map data of the current position and constructing a local three-dimensional map containing navigation and obstacle avoidance information in real time based on the three-dimensional environmental map include: Obtain the current area's environmental image data through the depth camera; According to the image depth information in the environmental map data, an obstacle information map containing navigation and obstacle avoidance information of the current area is constructed. The obstacle information map is integrated with the three-dimensional environment map to obtain a local three-dimensional map.

6. The method for mapping and navigation based on SLAM according to claim 4, characterized in that: Based on the visual SLAM algorithm, in the step of constructing an obstacle information map containing navigation and obstacle avoidance information of the current area in real time through the environmental map data: The navigation obstacle avoidance information includes environmental objects, obstacles, and passable area information.

7. The method for mapping and navigation based on SLAM according to claim 1, characterized in that: In the step of generating a navigation path using a path planning algorithm according to the exploration priority: The path planning algorithm adopts the Dijkstra path planning algorithm.

8. The method for mapping and navigation based on SLAM according to claim 1, characterized in that: The step of using a path planning algorithm to generate a navigation path according to the exploration priority also includes the following steps: Display the generated navigation path.

9. A mobile measurement backpack, characterized in that: The method for mapping and navigation based on SLAM as described in any one of claims 1 to 8 is applied, wherein the mobile measurement backpack comprises: A support frame, to which a backpack body is connected, and the backpack body is used to be placed on the back of an operator; A laser radar and a visual sensor, wherein the laser radar and the visual sensor are arranged on the supporting frame; A display controller is connected to the support frame at a side away from the backpack body and is used to be arranged in front of the operator's head.

10. The mobile measurement backpack according to claim 9, characterized in that: The laser radar and the visual sensor are combined into a measurement data acquisition module, and the measurement data acquisition module is detachably connected to the support frame; The display controller is movably connected to the support frame through a front frame, and the display controller can be flipped in the up and down direction by bending.

Images