Multi-sensor fusion transparent obstacle environment autonomous mapping method and system

Abstract

This invention relates to the field of mobile robot technology, and provides a method and system for autonomous mapping of transparent obstacle environments using multi-sensor fusion. The method includes: acquiring point cloud data of obstacles collected by a lidar sensor and obstacle distances collected by two other sensors; obtaining the obstacle configuration based on the obstacle distances collected by the two sensors; randomly selecting an obstacle distance collected by one sensor and calculating the difference between this distance and the average point cloud data; determining whether the obstacle is transparent based on the difference; if it is transparent, replacing the obstacle's point cloud data with the obstacle configuration after coordinate transformation; updating the grid map based on the obstacle's point cloud data and determining boundary points; and finding the optimal boundary point when the number of boundary points reaches a certain value to achieve autonomous mapping by the robot. This method can obtain an accurate configuration and fuse it with environmental information obtained from the lidar sensor to complete autonomous mapping in transparent obstacle environments.

Description

Technical Field

[0001] This invention belongs to the field of mobile robot technology, and in particular relates to a method and system for autonomous mapping of transparent obstacle environments using multi-sensor fusion. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] With the rapid development of robotics technology, various types of mobile robots are playing a significant role in emergency rescue, home services, and environmental exploration, leading to increasingly demanding positioning requirements. In indoor environments, building walls and underground spaces make it difficult for mobile robots to receive sufficient GPS signals for accurate positioning. The emergence of Simultaneous Localization and Mapping (SLAM) enables mobile robots to achieve localization and map creation in unknown indoor environments.

[0004] Various mobile robots perform tasks such as security patrols, item delivery, and disinfection spraying in indoor environments. When performing these tasks, they typically need to build a complete map of their surroundings during movement. However, environments like schools, office buildings, and shopping malls are filled with transparent obstacles such as glass curtain walls, glass doors, and glass fences, posing a challenge to the robots' safety. If the map does not indicate the corresponding locations of glass obstacles, it can lead to collisions with glass or even falls to the other side during navigation. Vision or laser beams directly acquire information about objects on the other side of the glass, so neither cameras nor lidar sensors can effectively detect glass in indoor environments. Ultrasonic sensors, a type of acoustic sensor, calculate the distance to obstacles by emitting ultrasonic waves and calculating the time it takes to receive the echo. Unaffected by the transparency of objects, they can be used to detect glass.

[0005] For the autonomous mapping task in the aforementioned transparent obstacle environment, other proposed ultrasonic sensor information fusion schemes have many problems, such as: using only one ultrasonic sensor for measurement, such as... Figure 5 As shown, the uncertainty of ultrasonic sensor measurements makes it impossible to obtain the accurate configuration of the obstacle; while two sensors can be used to measure the obstacle configuration, there are strict requirements on the installation position of the sensors.

[0006] Autonomous exploration and mapping of unknown areas by mobile robots is crucial for their ability to operate independently of human control. A commonly used method for autonomous exploration is the boundary point approach, which involves finding and traversing the boundary between known and unknown areas. The strategy for finding and selecting these boundary points is particularly important. The Rapidly-exploring Random Tree (RRT) algorithm probabilistically fills in regions in space to reach the boundary, and is frequently used in autonomous exploration to find boundary points. However, traditional RRT mapping suffers from oscillations, repeatedly traversing known areas. This is mainly due to insufficient alternative points, forcing the selection of distant boundary points, and the need to replan the path when a better boundary point is encountered during the movement. Summary of the Invention

[0007] To address the technical problems mentioned above, this invention provides a method and system for autonomous mapping of transparent obstacle environments using multi-sensor fusion. It uses two ultrasonic sensors to detect and construct transparent obstacles, obtaining accurate configurations and fusing them with environmental information obtained from lidar to complete autonomous mapping of transparent obstacle environments.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] The first aspect of the present invention provides a method for autonomous mapping of transparent obstacle environments using multi-sensor fusion, comprising:

[0010] Acquire point cloud data of obstacles collected by lidar, and distances to obstacles collected by the two sensors;

[0011] Based on the obstacle distances collected by two sensors, the obstacle configuration is obtained. The obstacle distance collected by one sensor is randomly selected, and the difference between the distance and the average point cloud data is calculated. Based on the difference, it is determined whether the obstacle is transparent.

[0012] If the obstacle is transparent, the obstacle configuration will be transformed in coordinates and the point cloud data of the obstacle will be replaced.

[0013] Based on the point cloud data of obstacles, the grid map is updated and the boundary points are determined. When the number of boundary points reaches a certain value, the optimal boundary point is found to achieve autonomous mapping by the robot.

[0014] Furthermore, the linear equations of the obstacles in the obstacle configuration are as follows:

[0015] y = kx + b

[0016] Among them, parameters If we take the robot's orientation relative to the two sensors as the y-axis, and determine the x-axis perpendicular to the y-axis using the right-hand rule, the first sensor is located m meters away from the origin in the negative x-axis direction, and the second sensor is located n meters away from the origin in the positive y-axis direction. The second sensor is oriented in the positive y-axis direction, and there is an angle between the first sensor's orientation and the positive half-axis of the y-axis. The obstacle distance collected by the first sensor is D. S1 The distance to the obstacle collected by the second sensor is D. S2 .

[0017] Furthermore, the start and end points in the obstacle configuration are the tangent points obtained by the straight line of the obstacle being tangent to the circumference of the two fan-shaped regions detected by the two sensors.

[0018] Furthermore, the length of the line segment in the obstacle configuration is:

[0019]

[0020] or,

[0021]

[0022] In this system, the robot's orientation relative to the two sensors is taken as the y-axis. The x-axis, perpendicular to the y-axis, is determined according to the right-hand rule. The first sensor is located at a distance of m meters from the origin in the negative x-axis direction, and the second sensor is located at a distance of n meters from the origin in the positive y-axis direction. The second sensor is oriented in the positive y-axis direction, and there is an angle between the orientation of the first sensor and the positive half-axis of the y-axis. α is the absolute value of the angle between the perpendicular line of the obstacle's line and the positive y-axis direction.

[0023] Furthermore, the coordinate transformation is as follows:

[0024]

[0025] Where ρ is the radius of the point cloud's position from the origin in the point cloud coordinate system, θ is the azimuth angle of the point cloud's position, and k and b are the parameters of the obstacle's straight-line equation.

[0026] Furthermore, the specific method for determining the boundary points is as follows:

[0027] Using the RRT algorithm, growth begins from an empty grid cell. During the growth process, new nodes are evaluated. If a new node is in an unknown grid cell and the line connecting it to its parent node crosses both an empty grid cell and an unknown grid cell, then the new node is defined as a boundary point.

[0028] Furthermore, a grid map is built and updated using the SLAM algorithm, and obstacles are represented in probabilistic form.

[0029] A second aspect of the present invention provides an autonomous mapping system for transparent obstacle environments based on multi-sensor fusion, comprising:

[0030] The data acquisition module is configured to acquire point cloud data of obstacles collected by the lidar and distances to obstacles collected by the two sensors.

[0031] The obstacle detection module is configured to: obtain the obstacle configuration based on the obstacle distances collected by two sensors, randomly select the obstacle distance collected by one sensor, calculate the difference between the distance and the average point cloud data, and determine whether the obstacle is transparent based on the difference.

[0032] The coordinate transformation module is configured to: if the obstacle is transparent, perform coordinate transformation on the obstacle configuration and replace the point cloud data of the obstacle;

[0033] The mapping module is configured to update the grid map based on the point cloud data of obstacles, determine the boundary points, and find the optimal boundary points when the number of boundary points reaches a certain value to achieve autonomous mapping by the robot.

[0034] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the multi-sensor fusion autonomous mapping method for transparent obstacle environments as described above.

[0035] A fourth aspect of the present invention provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the multi-sensor fusion autonomous mapping method for transparent obstacle environments as described above.

[0036] Compared with the prior art, the beneficial effects of the present invention are:

[0037] This invention provides a multi-sensor fusion method for autonomous mapping of transparent obstacle environments. It can obtain the accurate configuration of obstacles through two sensors and achieve better mapping results compared to other methods that use a single sensor for detection.

[0038] This invention provides a method for autonomous mapping of transparent obstacle environments using multi-sensor fusion. It reduces the uncertainty of ultrasonic sensor measurements through a simple algorithm, eliminating the need for complex operations such as transformation and feature extraction, thereby improving the efficiency, versatility, and ease of use of transparent obstacle detection algorithms.

[0039] This invention provides a method for autonomous mapping of transparent obstacle environments using multi-sensor fusion. To address the oscillation phenomenon during the exploration process, this invention proposes an improved boundary point selection strategy using an execution task pool, which, combined with information fused from multiple sensors, completes the autonomous mapping of transparent obstacle environments. Attached Figure Description

[0040] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0041] Figure 1 This is a flowchart of the multi-sensor fusion method for autonomous mapping of transparent obstacle environments according to Embodiment 1 of the present invention;

[0042] Figure 2 This is a top view of the vehicle assembly according to Embodiment 1 of the present invention;

[0043] Figure 3 This is a flowchart illustrating the boundary point selection logic of Embodiment 1 of the present invention.

[0044] Figure 4 This is a diagram of the lidar measurement model according to Embodiment 1 of the present invention;

[0045] Figure 5 This is a diagram of the ultrasonic sensor measurement model according to Embodiment 1 of the present invention;

[0046] Figure 6 This is a diagram of the obstacle detection configuration according to Embodiment 1 of the present invention. Detailed Implementation

[0047] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0048] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0049] Example 1

[0050] This embodiment provides a multi-sensor fusion method for autonomous mapping of transparent obstacle environments. It uses the Gmapping algorithm based on LiDAR and the RRT algorithm to find boundary points for mapping. During the mapping process, ultrasonic sensors are used to detect glass and add it to the corresponding parts of the map.

[0051] The multi-sensor fusion method for autonomous mapping of transparent obstacle environments provided in this embodiment uses two ultrasonic sensors to detect and construct transparent obstacles, which can obtain accurate configurations and fuse them with environmental information obtained from lidar. The mobile robot moves to the next target point according to the selected strategy, and at the same time uses the fused point cloud information to perform mapping work, and finally completes the environmental exploration.

[0052] like Figure 2As shown, both ultrasonic sensors and lidar are mounted on the mobile robot (vehicle). Three ultrasonic sensors are installed at a 30° angle at a certain distance in front of the vehicle, with two sensors forming a group. The sensors on the left and right sides share the sensor directly in front of the vehicle to detect obstacles in front of it. The lidar is installed on top of the vehicle in an unobstructed position to detect obstacles around the vehicle.

[0053] like Figure 1 As shown, the specific steps include the following:

[0054] Step 1: Information is collected by lidar and ultrasonic sensors.

[0055] LiDAR collects structural information about the surrounding environment, forming images such as... Figure 4 The point cloud data of some obstacles within a 270° range shown is given by point cloud information D = {d i |i=1,2,...,N}, where d i For each laser beam, the distance to the obstacle at the corresponding angle is given. Taking the single-line lidar used in this embodiment as an example, the lidar generates 1081 points per revolution, i.e., i = 1, 2, ..., 1081, evenly distributed within an effective 270° range. There are 4 laser point cloud information points within each 1°. Taking the area directly behind the lidar as 0° and increasing counterclockwise, the relationship between the point cloud index i and the angle θ is approximately assumed to be...

[0056] Meanwhile, the three ultrasonic sensors obtain the real-time distances to obstacles ahead, respectively, D. S1 D S2 D S3 Amplitude filtering is applied to the ultrasonic data to avoid instantaneous jump errors caused by factors such as vehicle movement and acoustic interference. The measured ultrasonic data is saved as a time series; to avoid excessive memory usage and to account for the stabilization process after data jumps, data from the current moment to six moments ago is saved. The ultrasonic sensor returns the measured distance D every 100ms. S1 D S2 Or D S3 The program creates an array Temp of length 7 to store the data received by a single sensor each time. When data is received, the elements in the array are shifted forward sequentially, i.e., Temp[i] = Temp[i+1], and the new data at the current moment is stored at the end of the array, Temp[6]. If the difference between the current data (the data in Temp[6]) and the initial data (Temp[0]) during this period is greater than a set threshold, the measurement data D will be discarded. S1 D S2 Or D S3 .

[0057] Step 2, Tangential configuration of the glass obstacle.

[0058] Figure 5 The image shows a measurement model for ultrasound. In this model, the ultrasound wave is simplified to a fan-shaped detection area with a certain range. The wavefront is an arc that spreads outwards. Regardless of where the obstacle is first encountered, that arc will be considered the obstacle's configuration. Figure 5 In the diagram, the black circle represents the location of the ultrasonic sensor, ω is the detection range angle, the shaded rectangle represents the location of obstacles in front, the fan-shaped area enclosed by the dashed line represents the unobstructed space in front of the sensor, and the radius R of the circle containing the fan-shaped area represents the distance D between the ultrasonic sensor and the obstacle. S1 D S2 Or D S3 .

[0059] Due to the uncertainty in detecting the obstacle configuration using a single ultrasonic sensor, two ultrasonic sensors are used for detection. The principle is as follows: Figure 6 As shown, after the wavefronts of two ultrasonic waves propagate forward, contact the obstacle, and return, data from the two sensors are obtained. Let the vehicle's orientation be the y-axis. In the top-view plane, the x-axis is determined perpendicular to the y-axis according to the right-hand rule. Ultrasonic sensor S1 is located m meters away from the origin in the negative x-axis direction, and sensor S2 is located n meters away from the origin in the positive y-axis direction. S2 is oriented towards the positive y-axis, and the angle between the orientation of sensor S1 and the positive y-axis is 30°. The detected distance is D. S1 With D S2 Then the radii of the two sector regions are D. S1 and D S2 According to the measurement model of ultrasound, the obstacle lies on the common tangent of the wavefronts of the two sound waves, and the equation of the tangent is:

[0060] y = kx + b

[0061] Among them, parameters After obtaining the linear equation of the obstacle, the points of tangency between this line and the circles containing the two sectors are used as the start and end points of the obstacle configuration. A line segment is then defined as the obstacle, and the length L of this line segment is:

[0062]

[0063] or,

[0064]

[0065] In the formula, α is the absolute value of the angle between the perpendicular line of the obstacle segment and the positive direction of the y-axis.

[0066] Step 3: Compare Information to Determine the Glass. Due to the light transmittance of glass, lidar detection of glass typically yields distance data behind the glass, resulting in a difference between this and the data from the ultrasonic sensor. Based on the actual geometric positional relationship between the sensors, the ultrasonic data is compared with the lidar data at the corresponding angle. Specifically, the program calculates the obstacle line segment configuration based on a set (two) of sensor data. The position of the endpoint of the line segment in the polar coordinate system is the corresponding angle from which the lidar data was obtained. Considering the fan-shaped detection range of the ultrasonic sensor, the lidar data at a certain angle corresponding to a specific ultrasonic sensor is averaged over a range and used as the lidar data at that angle. Assume a sensor data D... S1 The average point cloud data of the lidar at the corresponding location The difference in That is, M laser point cloud data points near the corresponding angle of the endpoint are selected and averaged. In this embodiment, M is set to 20. The difference is used to determine whether it is a transparent obstacle. Specifically, if the difference Δ is greater than the set value T, the obstacle is a transparent obstacle (glass); otherwise, the obstacle is a normal obstacle. The judgment rule is as follows:

[0067]

[0068] Where T is the threshold, which is determined through actual experimental procedures.

[0069] If the judgment result is "glass", the point cloud will be replaced with the configuration obtained by ultrasonic detection within a certain range.

[0070] The obtained line segment is transformed to replace the corresponding LiDAR point cloud. The equation of the line in the point cloud coordinate system (polar coordinate system) is:

[0071]

[0072] In the formula, ρ is the radius of the point cloud position from the origin in the point cloud coordinate system, and θ is the azimuth angle of the point cloud position. In this method, the starting position of the azimuth angle is 0°, that is, the polar axis is located in the positive x-axis direction in the rectangular coordinate system, and it increases in the counterclockwise direction.

[0073] Step 4: Obtain the initial location and map, and use RRT to explore the boundaries.

[0074] After obtaining several frames of point cloud data, an updated occupancy grid map is acquired using the SLAM algorithm (Gmapping algorithm in this embodiment). The grid map represents obstacles in the real-world environment in probabilistic form. The resulting map includes "free," "occupied," and "unknown" states. Unobserved "unknown" grids are initially represented with an occupancy probability of 0.5. This probability is updated based on subsequent observations using various sensors. "Free" and "occupied" represent the already explored areas; "free" grids have an occupancy probability of 0, while "occupied" grids have an occupancy probability of 1. Let X represent all grids in the two-dimensional space, then X... free Indicates an empty grid cell, X occ This indicates that an obstacle exists when a grid cell is occupied, X unk Indicates an unknown raster.

[0075] Using the RRT algorithm, from X free Growth begins, and during the growth process, new nodes are evaluated. If the new node is in X... unk The line connecting the middle node and its parent node crosses the X-axis. free With X unk If so, then that node is defined as a boundary point. Taking local RRT as an example, the root node of the random tree is generated at the location of the robot in Xf. ree From here, it grows randomly. If it starts from X... free The new node obtained by growing the node in X is located at X. free If the new node is in X, then the random tree continues to grow. unk If the new node is considered a boundary point, it is retained while the entire local random tree is deleted. By setting a small growth step size, the new node can be approximated as being on the boundary.

[0076] Step 5: Obtain the boundary points and use a selection strategy to select the target boundary points.

[0077] To avoid the robot repeatedly traversing the same areas due to constant changes in target boundary points during exploration, this method establishes a task pool to store candidate boundary points. When the number of boundary points has not reached a certain threshold, the robot waits in place, allowing the random tree to grow and find a sufficient number of boundary points to store in the task pool. This ensures a sufficient number of candidate boundary points to find the optimal boundary point for exploration. While waiting, the LiDAR and sensors continuously acquire environmental information. Once the threshold is reached, the reward for each point in the pool is calculated. The reward is calculated by subtracting the Euclidean distance C from the number of unknown grid cells I within a certain radius R of the boundary point from the Euclidean distance C from the robot to the boundary point. I and C are multiplied by a certain proportional coefficient in the calculation, where the certain radius R is determined by the coverage area of ​​the sensors on the robot. The optimal boundary point, i.e., the boundary point with the highest reward, is found. The robot moves to the optimal boundary point to obtain unknown environmental information around the boundary point for mapping. During the movement, non-boundary points in the pool are deleted, and points in the boundary point queue are added to the pool in the order of discovery. This process is repeated until all boundary points have been explored. The algorithm flow is as follows: Figure 3 As shown, points in the pool's outer boundary point queue will not participate in the profit calculation. This ensures that newly generated boundary points will not interfere with the ongoing movement process, thus avoiding oscillation problems.

[0078] The robot will then repeat the above process until the task pool is emptied, which signifies the completion of the exploration.

[0079] In this embodiment, the vehicle is a four-wheel differential car capable of forward, reverse, and differential rotation. At the highest point of the vehicle is a single-line lidar, model HOKUYO UST-10LX, with a radar scanning frequency of 40Hz, a scanning range of 270°, a ranging range of 0.06-10m, and 1081 scan points per cycle. Three ultrasonic sensors are mounted at the front of the vehicle, their installation positions and orientations as shown... Figure 1 As shown, the ultrasonic ranging range is 0.03-3m, and the detection range angle is 45°±15°.

[0080] The vehicle control hardware consists of an industrial computer and a low-level controller. The industrial computer used in the vehicle is a ROScube PicoTGL, RQP-T35 with an Intel Core i5-1145G7E processor. The low-level controller uses an STM32F103 development board. The industrial computer runs Ubuntu 20.04 under Linux, using the ROS environment. It is responsible for subscribing to and fusing data from various sensors, performing SLAM mapping using the Gmapping algorithm, finding boundary points using the RRT algorithm, and planning routes and issuing movement commands in the current map environment using ROS Move Base. The low-level controller is responsible for receiving movement commands from the industrial computer and driving the vehicle's motors to complete the movement.

[0081] In the ROS system, ultrasonic and lidar data are received, obstacles are constructed at corresponding locations, and the lidar data is evaluated and replaced, completing the sensor data fusion. Specifically, during the autonomous mapping process while the vehicle is moving, lidar and ultrasonic sensors continuously collect data and compare it with data from the same point in time. When the distance difference between the ultrasonic sensor and the corresponding lidar location reaches or exceeds a threshold, the line segment obstacle configuration formed by the two ultrasonic sensors at that location is replaced in the lidar data at the corresponding location. The fused data will be published in a new topic / Scan_replacesd with the lidar topic structure. Its content includes: basic data information Std_msgs / Headerheader, data start angle float32 angle_min, data end angle float32 angle_max, data angle minimum increment float32 angle_increment, data time minimum increment float32 time_increment, data acquisition time float32 scan_time, minimum distance float32 range_min, maximum distance float32 range_max, distance array float32[]ranges for each angle, and intensity array float32[]intensities for each angle.

[0082] The Gmapping algorithm subscribes to the ` / Scan_replacesd` topic to construct a map of the current exploration area. After obtaining the first frame of the map, the RRT algorithm is used to find boundary points. The RRT algorithm uses two random trees to find boundary points: a global tree and a local tree. The global tree starts at the robot's initial location and grows continuously throughout the entire map to find boundary points, ensuring that the entire map can be explored. The local tree also starts at the robot's location, but it is deleted after finding a boundary point and regenerates at the robot's current location, which helps the robot quickly obtain surrounding boundary points to complete the exploration. All found boundary points are added to the task pool and filtered according to a strategy to obtain the optimal boundary point for the current stage for exploration. After obtaining the target boundary point, the MoveBase function package issues a movement command to the target point to the underlying control board. Considering factors such as vehicle performance and sensor performance, the vehicle's movement speed is limited to less than 0.6 m / s.

[0083] This embodiment provides a multi-sensor fusion-based autonomous mapping method for transparent obstacle environments, offering corresponding calculation formulas for different installation locations. Considering practical usage requirements and vehicle size, sensor orientation and position can be more flexible. Using two sensors, the accurate configuration of the obstacle can be obtained, achieving better mapping results compared to detection with a single sensor. Furthermore, a simple algorithm reduces the uncertainty of ultrasonic sensor measurements, eliminating the need for complex operations such as transformation and feature extraction, thus improving the efficiency, versatility, and ease of use of the transparent obstacle detection algorithm. For oscillation phenomena during exploration, this invention proposes an improved boundary point selection strategy for the execution task pool, combining information from multi-sensor fusion to complete autonomous mapping in transparent obstacle environments.

[0084] Example 2

[0085] This embodiment provides an autonomous mapping system for transparent obstacle environments based on multi-sensor fusion, which specifically includes:

[0086] The data acquisition module is configured to acquire point cloud data of obstacles collected by the lidar and distances to obstacles collected by the two sensors.

[0087] The obstacle detection module is configured to: obtain the obstacle configuration based on the obstacle distances collected by two sensors, randomly select the obstacle distance collected by one sensor, calculate the difference between the distance and the average point cloud data, and determine whether the obstacle is transparent based on the difference.

[0088] The coordinate transformation module is configured to: if the obstacle is transparent, perform coordinate transformation on the obstacle configuration and replace the point cloud data of the obstacle;

[0089] The mapping module is configured to: update the raster map based on the point cloud data of obstacles, determine boundary points, and when the number of boundary points reaches a certain value, find the optimal boundary points to achieve autonomous mapping by the robot.

[0090] It should be noted that each module in this embodiment corresponds one-to-one with each step in Embodiment 1, and their specific implementation processes are the same, so they will not be repeated here.

[0091] Example 3

[0092] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the multi-sensor fusion autonomous mapping method for transparent obstacle environments as described in Embodiment 1 above.

[0093] Example 4

[0094] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the multi-sensor fusion autonomous mapping method for transparent obstacle environments as described in Embodiment 1 above.

[0095] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0096] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0100] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for autonomous mapping of transparent obstacle environment using multi-sensor fusion, characterized in that, The method comprises the following steps: acquiring point cloud data of an obstacle collected by a laser radar and obstacle distances collected by two ultrasonic sensors; obtaining an obstacle configuration based on the obstacle distances collected by the two sensors, randomly selecting an obstacle distance collected by one sensor, calculating a difference value with average point cloud data, and judging whether the obstacle is transparent based on the difference value; if the obstacle is transparent, replacing the point cloud data of the obstacle after coordinate transformation of the obstacle configuration; updating a grid map based on the point cloud data of the obstacle and determining boundary points, and when the number of the boundary points reaches a certain value, finding optimal boundary points to realize autonomous mapping of the robot; a straight line equation of the obstacle in the obstacle configuration is: Wherein, the parameters ; ; if the robot direction of the two sensors is the y-axis, the x-axis perpendicular to the y-axis is determined according to the right-hand rule, the first sensor is located at a position of -d1 meters in the negative direction of the x-axis of the coordinate origin, the second sensor is located at a position of d2 meters in the positive direction of the y-axis of the coordinate origin, the second sensor faces the positive direction of the y-axis, and the first sensor faces an angle with the positive half-axis of the y-axis. m n ;​​​ start and end points in the obstacle configuration are tangent points obtained by the tangent of a straight line of the obstacle and a circumference of two fan-shaped regions detected by the two sensors.

2. The multi-sensor fused transparent barrier environment autonomous mapping method of claim 1, wherein, a length of a line segment in the obstacle configuration is: or Wherein, the robot toward which the two sensors are located is y axis, x axis perpendicular to y axis is determined according to right-hand rule, the first sensor is located at the position of the negative direction of x axis of the coordinate origin at a distance of 0.5 meters m The second sensor is located at the position of the positive direction of y axis of the coordinate origin at a distance of 0.5 meters n The second sensor is toward the positive direction of y axis, the first sensor is toward the positive half axis of y axis, and there is an included angle between the first sensor and the positive direction of y axis, The absolute value of the angle between the perpendicular line of the straight line of the obstacle and the positive direction of y axis.

3. The multi-sensor fused transparent barrier environment autonomous mapping method of claim 1, wherein, the coordinate transformation is: wherein, is a radius length of a position where the point cloud is located in a point cloud coordinate system from an origin, is an azimuth angle of the position where the point cloud is located, k and b are parameters of a straight line equation of the obstacle.

4. A multi-sensor fused transparent obstacle environment autonomous mapping system, characterized in that, The method comprises the following steps: a data acquisition module configured to acquire point cloud data of an obstacle collected by a laser radar and obstacle distances collected by two ultrasonic sensors; an obstacle judgment module configured to obtain an obstacle configuration based on the obstacle distances collected by the two sensors, randomly select an obstacle distance collected by one sensor, calculate a difference value with average point cloud data, and judge whether the obstacle is transparent based on the difference value; a straight line equation of the obstacle in the obstacle configuration is: Wherein, the parameters ; ; if the orientation of the robot where the two sensors are located is the y-axis, the x-axis perpendicular to the y-axis is determined according to the right-hand rule, the first sensor is located at a position of -d1 meters in the negative direction of the x-axis of the coordinate origin, the second sensor is located at a position of d2 meters in the positive direction of the y-axis of the coordinate origin, the second sensor faces the positive direction of the y-axis, and the first sensor faces an angle with the positive half-axis of the y-axis. m n ;​​​ start and end points in the obstacle configuration are tangent points obtained by the tangent of a straight line of the obstacle and a circumference of two fan-shaped regions detected by the two sensors. a coordinate transformation module configured to replace the point cloud data of the obstacle after coordinate transformation of the obstacle configuration if the obstacle is transparent; a mapping module configured to update a grid map based on the point cloud data of the obstacle and determine boundary points, and when the number of the boundary points reaches a certain value, find optimal boundary points to realize autonomous mapping of the robot.

5. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to realize the steps in the multi-sensor fusion autonomous mapping method of a transparent obstacle environment according to any one of claims 1-3.

6. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to realize the steps in the multi-sensor fusion autonomous mapping method of a transparent obstacle environment according to any one of claims 1-3.

Images