Robot autonomous positioning method, device and medium based on adaptive extension reset

Abstract

The application discloses a robot autonomous positioning method and device based on adaptive expansion reset and a medium, belongs to the technical field of artificial intelligence, and solves the problems of high calculation complexity and low real-time adaptability in the robot positioning process.The application firstly collects single-line laser radar, odometer and two-dimensional grid map data, and initializes a particle state set; then matches the single-line laser radar data with the two-dimensional grid map, calculates the likelihood value of each particle and a global matching index, calculates an adaptive factor when the global matching index is smaller than a set threshold, calculates expansion reset parameters according to the adaptive factor, and performs expansion reset on each particle; finally, the particle set after the expansion reset is used to update the particle weight, and the current pose of the robot is obtained; the application can effectively solve the problems of particle degradation and low real-time adaptability in the robot positioning process, and improve the robustness of the robot positioning.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology and relates to a robot autonomous localization method, device and medium based on adaptive extended reset. Background Technology

[0002] With the development of robot intelligence and industrial applications, real-time robot localization plays a crucial role. Whether it's an outdoor inspection robot or an indoor service robot, real-time localization is indispensable, especially in indoor environments facing dynamic obstacles and congestion, where robot navigation often results in position loss and motion planning failures. Therefore, providing a localization method that is robust and computationally efficient for indoor environments is essential.

[0003] Most existing indoor service robots rely on Monte Carlo pose estimation methods for global localization, which are based on particle filters to estimate robot positions. Existing technologies, such as the invention patent with publication number CN110986956A, disclose an autonomous learning global localization method based on an improved Monte Carlo algorithm. This method creates a particle filter, reads the current laser and a known map, and obtains the optimal initial pose through a scan-matching formula. This technology improves the overall accuracy of the particle swarm by incorporating scan-matching and also adds the function of automatically providing the initial pose, replacing manual pose assignment via a host computer and simplifying the operation. However, since each particle carries pose and weight information, resampling of all particles is required during localization, leading to a large number of particles failing to converge in real time, resulting in high computational complexity. Furthermore, because the original Monte Carlo algorithm adapts to a fixed number of particles, it cannot adaptively recover after localization failures. Summary of the Invention

[0004] The technical solution of this invention is used to solve the problems of high computational complexity and low real-time performance of adaptive recovery in robot localization.

[0005] The present invention solves the above-mentioned technical problems through the following technical solutions:

[0006] The robot autonomous localization method based on adaptive extended reset includes the following steps:

[0007] S1, Install a single-line lidar on the differential robot to acquire single-line lidar data, odometer data and two-dimensional grid map;

[0008] S2, Initialize the particle set, where each particle contains a pose and weights;

[0009] S3 updates motion based on odometry data, updates particle pose, and calculates the likelihood between single-line lidar data and two-dimensional grid map.

[0010] S4, calculate the new particle weights based on the likelihood;

[0011] S5. Compare the new particle weight with the preset threshold. If the new particle weight is less than the preset threshold, execute S6 to perform adaptive expansion reset resampling; if the new particle weight is greater than or equal to the preset threshold, execute S8 to perform regular resampling.

[0012] S6. Calculate the adaptive factor based on the new particle weights, and then calculate the extended parameters based on the adaptive factor.

[0013] S7, adaptively expand, reset and update the weights of each particle, then execute S9;

[0014] S8, perform routine resampling based on cumulative weight and uniform sampling interval, then execute S9;

[0015] S9. Based on the pose of each sampled particle, calculate the weighted average pose to obtain the final pose estimate of the differential robot.

[0016] S10: When new single-line lidar data is received, process the new data in a loop and return to execute S3.

[0017] Furthermore, the single-line lidar data mentioned in S1 is obtained by acquiring single-line lidar installed on the differential robot, the odometer data is obtained based on the kinematic model of the differential robot, and the two-dimensional grid map is known data acquired in advance.

[0018] Furthermore, S2 specifically involves: initializing the particle set P = {(p i ,w i )},i∈N, where N represents the number of particles, p i Let p represent the pose of the i-th particle. i =(x i ,y i ,θ i ), x i Let y represent the distance of the i-th particle in the x-direction, and y represent the distance of the i-th particle i θ represents the distance of the i-th particle in the y-direction. i w represents the direction angle of the i-th particle; i The weight of the i-th particle is denoted as .

[0019] Further, S3 includes:

[0020] S31, perform motion update based on odometry data to obtain the pose p of the i-th particle at time t.i_t =(x i_t ,y i_t ,θ i_t ), where x i_t The distance of the i-th particle in the x-direction at time t is represented by y. i_t θ represents the distance of the i-th particle in the y-direction at time t. i_t This represents the orientation angle of the i-th particle at time t.

[0021] S32, Calculate the likelihood L(p) between the single-line lidar data and the two-dimensional grid map. i Specifically:

[0022] Based on the current i-th particle pose p at time t i_t =(x i_t ,y i_t ,θ i_t The single-line lidar data is converted onto a two-dimensional grid map, and the coordinates of the k-th single-line lidar beam converted onto the two-dimensional grid map are calculated. Using the following logical representation:

[0023]

[0024]

[0025] in, Let be the distance measurement value of the k-th single-line lidar beam at time t. This represents the angle measurement value of the k-th single-line lidar beam. These are the x and y coordinates of the k-th single-line lidar beam converted to a two-dimensional grid map, respectively.

[0026] Based on the improved likelihood calculation formula, the likelihood L(p) between single-line lidar data and a two-dimensional grid map is calculated using the following logic. i ):

[0027]

[0028] Wherein d(p i ) represents the coordinates The distance to the nearest obstacle, where σ is a normalized parameter.

[0029] Furthermore, in S4, the new particle weight w is calculated using the following logic. i_new :

[0030]

[0031] Wherein, L(p i) represents the likelihood of single-line lidar data and a two-dimensional grid map, and N represents the number of particles.

[0032] Further, S6 includes:

[0033] S61, calculate the adaptive factor λ using the following logical representation:

[0034]

[0035] Where, α th w is a preset threshold. i_new For the new particle weights;

[0036] S62, For each particle p that needs to be reset i =(x i ,y i ,θ i The extended parameters are calculated based on the adaptive factor λ; the extended parameters include the position extension r. e With angle extension Δθ max This can be represented using the following logic:

[0037] r e =r min +λ·(r max -r min )

[0038] Δθ max =θ min +λ·(θ max -θ min )

[0039] Where, r min and r max θ represents the minimum and maximum allowable expansion radius in the environment, respectively. min and θ max These represent the minimum and maximum permissible angular offsets, respectively.

[0040] Further, S7 includes:

[0041] S71, Calculate the particle set P = {(p_t)} at time t. i_t ,w i_t The covariance matrix of}, where w i_t The weight of the i-th particle at time t is represented by the following logic:

[0042]

[0043] Where, x i_t The distance of the i-th particle in the x-direction at time t is represented by y. i_tw represents the distance of the i-th particle in the y-direction at time t. i_t Let be the weight of the i-th particle at time t. Let be the average values ​​of the i-th particle in the x and y directions at time t, respectively.

[0044] Based on ∑, we perform eigenvalue decomposition to obtain eigenvalues ​​λ1 and λ2 and eigenvectors v1 and v2, satisfying λ1>λ2;

[0045] When λ1 < 0.1, reduce the resampling frequency; when λ1 > 0.1, disperse the particles and execute S72.

[0046] S72, perform extended sampling for each particle that needs to be reset, and obtain the sampled pose of each particle as p. i ′ =(x i ′ ,y i ′ ,θ i ′ The following logic can be used to represent this:

[0047]

[0048] in, and These represent the average value of the particle in the x-direction, the average value in the y-direction, and the y-direction value, respectively. This represents solving ∑w i_t sinθ i_t With ∑w i_t cosθ i_t The arctangent value, ξ x ξ y ξ θ All are random offsets of particle states. It meets the requirements of a mean of 0 and a variance of 0. Gaussian distribution, It meets the requirements of a mean of 0 and a variance of 0. The Gaussian distribution is obtained, and step S9 is executed.

[0049] Further, S8 specifically involves: calculating the cumulative weight w of the first to j particles. j_c With uniform sampling interval point u i This can be represented using the following logic:

[0050]

[0051] Among them, w j Let r be the weight of the j-th particle, j∈i, and r be a random number between 0 and 1 / N; for each ui Find the one that satisfies u i ≤w j_c At time j, obtain the pose of each particle after sampling, complete the regular resampling, and execute step S9.

[0052] An electronic device includes a memory and a processor, the memory being used to store a program that supports the processor in executing the above-described adaptive extended reset-based robot autonomous localization method, the processor being configured to execute the program stored in the memory.

[0053] A storage medium storing a computer program, which, when executed by a processor, performs the steps of the above-described robot autonomous localization method based on adaptive extended reset.

[0054] The advantages of this invention are:

[0055] (1) This invention first collects data from a single-line lidar, odometry, and a two-dimensional grid map, and initializes the particle state set. Then, it matches the single-line lidar data with the two-dimensional grid map, calculates the likelihood value of each particle and the global matching index. When the global matching index is less than a set threshold, it calculates an adaptive factor, calculates extended reset parameters based on the adaptive factor, and performs extended reset on each particle. Finally, it updates the particle weights using the extended reset particle set to obtain the robot's current pose. This invention can effectively solve the problems of particle degradation and low real-time performance in robot localization, and improve the robustness of robot localization.

[0056] (2) This invention modifies the likelihood calculation method of single-line lidar data and two-dimensional grid map, introduces an improved likelihood calculation formula, simplifies the likelihood calculation model, reduces invalid calculations, and compared with the point-by-point matching of traditional AMCL, it can reduce oversensitivity to low error regions, avoid repeated calculations, and achieve lower computational complexity.

[0057] (3) By introducing a new adaptive factor calculation method, this invention can dynamically calculate the adaptive factor and adjust the position and angle expansion parameters, thereby adjusting the particle expansion range in real time according to the positioning situation, adaptively shrinking or expanding the particle expansion range, and reducing the particle quantity requirement; it can realize adaptive updating of particle weights when the robot positioning is lost, quickly restore the robot positioning, effectively improve the real-time performance and anti-degradation ability of robot positioning, and solve the particle degradation problem.

[0058] (4) When the adaptive factor λ approaches 0, r e ≈r min , Δθ max ≈θ minAt this point, the expansion range is narrowed, and particles are concentrated in a high-probability region near the robot's true pose. This reduces redundant calculations and improves positioning accuracy. However, when the adaptive factor λ approaches 1, the positioning may have been lost, and the position expansion r... e ≈r max By expanding the particle search range to cover a wider area for localization recovery, and simultaneously extending the angle to the maximum allowable range, localization failure due to directional deviation is avoided. Therefore, by dynamically calculating the adaptive factor λ to dynamically adjust the expansion range, particles are avoided in low-probability areas, focusing computational resources. Expansion calculations are reduced during the convergence phase and precisely expanded during the divergence phase, reducing the overall computational load. This invention introduces expansion parameters to update particles for both position and angle expansion ranges, enabling rapid robot localization calculations using a smaller number of particles, thus improving the overall robot localization efficiency. Attached Figure Description

[0059] Figure 1 This is a flowchart of the robot autonomous localization method based on adaptive extended reset according to Embodiment 1 of the present invention;

[0060] Figure 2 This is a structural block diagram of the robot autonomous localization process according to Embodiment 1 of the present invention;

[0061] Figure 3 This is an example diagram of the robot's autonomous localization at different times based on a visualization tool, according to Embodiment 1 of the present invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments:

[0064] Example 1

[0065] like Figure 2As shown, to implement a robot autonomous localization method based on adaptive expansion reset, three coordinate systems must be transformed: the base coordinate system, the odometry coordinate system, and the map coordinate system. The transformation between the base and odometry coordinate systems is analyzed by the kinematic model to obtain odometry data. However, the transformation between the odometry and map coordinate systems can accumulate errors due to long-term motion, resulting in incorrect localization results. This invention combines single-line LiDAR data, odometry data, and a two-dimensional grid map to provide a robot autonomous localization method based on adaptive expansion reset. By using adaptive expansion reset and dynamic adjustment of the particle count, the number of particles required for real-time particle updates during localization is reduced, lowering computational complexity, and enabling rapid recovery after localization failure.

[0066] like Figure 1 Specifically, a robot autonomous localization method based on adaptive extended reset is disclosed, including the following steps:

[0067] Step S1: Install the single-line lidar on the differential robot, collect single-line lidar data, and obtain odometer data and a two-dimensional grid map.

[0068] In this embodiment, the single-line lidar data is acquired by a single-line lidar installed on the differential robot, the odometer data is obtained based on the kinematic model of the differential robot, and the two-dimensional grid map is known data acquired in advance.

[0069] In this embodiment, the Differential Drive Robot is a common type of mobile robot, characterized by movement controlled by two independently driven wheels (usually left and right). The kinematic model of the differential drive robot describes how the robot achieves translation, turning, and rotation through the difference in rotational speed between the left and right wheels. It forms the basis for robot navigation, path planning, and localization. The mathematical expression of its kinematic model is not explained here. In this embodiment, the two-dimensional grid map refers to a two-dimensional data structure displayed as an image using visualization algorithms, typically presented as a black-and-white or color image.

[0070] Step S2, initialize a set of N particles, denoted as P = {(p i ,w i )},i∈N, where N represents the number of particles, p i Let p represent the pose of the i-th particle. i =(x i ,y i ,θ i ), x i Let y represent the distance of the i-th particle in the x-direction, and y represent the distance of the i-th particle i θ represents the distance of the i-th particle in the y-direction.i w represents the direction angle of the i-th particle; i The weight of the i-th particle is denoted as .

[0071] Step S3 involves updating the motion based on the odometry data, updating the particle pose, and calculating the likelihood between the single-line lidar data and the two-dimensional grid map. Step S3 specifically includes the following steps:

[0072] S31, perform motion update based on odometry data to obtain the pose p of the i-th particle at time t. i_t =(x i_t ,y i_t ,θ i_t ), where x i_t The distance of the i-th particle in the x-direction is represented by y at time t. i_t θ represents the distance of the i-th particle in the y-direction at time t. i_t This represents the orientation angle of the i-th particle at time t.

[0073] S32, Calculate the likelihood L(p) between the single-line lidar data and the two-dimensional grid map. i Specifically:

[0074] First, based on the i-th particle pose p at time t... i_t =(x i_t ,y i_t ,θ i_t The single-line lidar data is converted onto a two-dimensional grid map, and the coordinates of the k-th single-line lidar beam converted onto the two-dimensional grid map are calculated. Using the following logical representation:

[0075]

[0076] in, Let be the distance measurement value of the k-th single-line lidar beam at time t. This represents the angle measurement value of the k-th single-line lidar beam. These are the x and y coordinates, respectively, of the k-th single-line lidar beam converted to a two-dimensional grid map.

[0077] Then, based on the improved likelihood calculation formula, the likelihood L(p) between the single-line lidar data and the two-dimensional grid map is calculated using the following logic. i ):

[0078]

[0079] Wherein d(p i ) represents the coordinates The nearest obstacle distance, σ is a normalization parameter, and in this embodiment, σ can be taken as twice the resolution of the two-dimensional raster map.

[0080] In contrast, the original likelihood L odom (p i The formula for calculating ) is:

[0081]

[0082] It is evident that the improved likelihood calculation formula provided in this embodiment can significantly reduce the complexity of likelihood calculation and improve computational efficiency.

[0083] Step S4: Calculate the new particle weight w based on the likelihood. i_new .

[0084] In this embodiment, the new particle weight w i_new Calculated based on the average likelihood of all particles, using the following logical representation:

[0085]

[0086] Wherein, L(p i ) represents the likelihood of single-line lidar data and a two-dimensional grid map, and N represents the number of particles.

[0087] In the traditional Adaptive Monte Carlo Localization (AMCL) algorithm, likelihood calculation is usually based on point-by-point matching of LiDAR data and map. However, this embodiment simplifies the likelihood calculation model by introducing an improved likelihood calculation formula, which can reduce invalid calculations. Compared with the point-by-point matching of traditional AMCL, it can reduce oversensitivity to low-error areas, avoid repeated calculations, and achieve lower computational complexity.

[0088] Step S5, compare the new particle weights w i_new With the preset threshold α th Size: when w i_new Less than α th When w is in use, step S6 is executed to perform adaptive extended reset resampling; i_new Greater than or equal to α th At this time, step S8 is executed to perform routine resampling. In this embodiment, a preset threshold α is used. th We can take α th ∈[0.5,0.9].

[0089] Step S6, based on the new particle weights w i_new Calculate the adaptive factor λ, and then calculate the extended parameters based on the adaptive factor λ; step S6 specifically includes the following steps:

[0090] S61. Calculate the adaptive factor λ using the following logical representation:

[0091]

[0092] Where, α th w is a preset threshold. i_new For the new particle weights.

[0093] S62, For each particle p that needs to be reset i =(x i ,y i ,θ i The extended parameters are calculated based on the adaptive factor λ; the extended parameters include the position extension r. e With angle extension Δθ max This can be represented using the following logic:

[0094] r e =r min +λ·(r max -r min )

[0095] Δθ max =θ min +λ·(θ max -θ min )

[0096] Where, r min and r max These represent the minimum and maximum permissible expansion radii in the environment, respectively, and are mainly related to the size of the environmental map. For small indoor maps, the default value is ±0.05 to ±0.1 meters, while for large outdoor maps, the value ranges from ±0.2 to ±0.5 meters. θ min and θ max These represent the minimum and maximum permissible angular offsets, respectively, with a default range of -0.2 to 0.2 rad.

[0097] In this embodiment, the position expansion radius (r) min and r max The angle expansion range (θ) is dynamically adjusted according to the resolution of the environmental map. min and θ max Based on the robot's maximum steering angle, the adaptive factor λ is dynamically calculated, and the extended parameter r is adjusted accordingly. e With Δθ max It can adjust the particle expansion range in real time according to the positioning situation, adaptively shrinking or expanding the particle expansion range, thus reducing the number of particles required. This effectively improves the real-time performance and anti-degradation capability of robot positioning and solves the particle degradation problem.

[0098] When the adaptive factor λ approaches 0, r e ≈r min , Δθmax ≈θ min At this point, the expansion range is narrowed, and particles are concentrated in a high-probability region near the robot's true pose. This reduces redundant calculations and improves positioning accuracy. However, when the adaptive factor λ approaches 1, the positioning may have been lost, and the position expansion r... e ≈r max By expanding the particle search range to cover a wider area to restore localization, and simultaneously extending the angle to the maximum allowable range, localization failure due to directional deviation is avoided. Therefore, by dynamically calculating the adaptive factor λ to dynamically adjust the expansion range, particles can be avoided in low-probability areas, computational resources can be focused, expansion calculations can be reduced during the convergence phase, and expansion can be precisely extended during the divergence phase, thus reducing the overall computational load.

[0099] Step S7, adaptively expand, reset, and update the weights of each particle; step S71 specifically includes the following steps:

[0100] S71, Calculate the particle set P = {(p_t)} at time t. i_t ,w i_t The covariance matrix of}, where w i_t This represents the weight of the i-th particle at time t. In this embodiment, the covariance matrix describes the particle distribution characteristics and is used to evaluate the particle positioning uncertainty. The covariance matrix ∑ is represented by the following logic:

[0101]

[0102] Among them, w i_t Let be the weight of the i-th particle at time t. Let be the average values ​​of the i-th particle in the x and y directions at time t, respectively.

[0103] Based on ∑, we can perform eigenvalue decomposition to obtain eigenvalues ​​λ1 and λ2 and eigenvectors v1 and v2, satisfying λ1>λ2.

[0104] When λ1 < 0.1, reduce the resampling frequency; when λ1 > 0.1, disperse the particles and execute S72.

[0105] S72, perform extended sampling for each particle that needs to be reset, and obtain the sampled pose of each particle as p. i ′ =(x i ′ ,y i ′ ,θ i ′ The following logic can be used to represent this:

[0106]

[0107] in, and These represent the average value of the particle in the x-direction, the average value in the y-direction, and the y-direction value, respectively. This represents solving ∑w i_t sinθ i_t With ∑w i_t cosθ i_t The arctangent value, ξ x ξ y ξ θ All are random offsets of particle states.

[0108] Furthermore, It meets the requirements of a mean of 0 and a variance of 0. Gaussian distribution, It meets the requirements of a mean of 0 and a variance of 0. The Gaussian distribution is obtained, and step S9 is executed.

[0109] Step S8: Perform regular resampling based on the cumulative weight and uniform sampling interval, and calculate the cumulative weight w of particles 1 to j. j_c With uniform sampling interval point u i This can be represented using the following logic:

[0110]

[0111] Where r is 0 to random number, w j For the weight of the j-th particle, w j_c The cumulative weight of particles 1 to j, u i Uniform sampling interval points; for each u i Find the one that satisfies u i ≤w j_c At time j, obtain the pose of each particle after sampling, complete the regular resampling, and execute step S9.

[0112] Step S9: Calculate the weighted average pose based on the sampled pose of each particle to obtain the robot's final pose estimate p. * =(x * ,y * ,θ * ).

[0113] Step S10: When new single-line lidar data is received, process the new data in a loop and return to step S3.

[0114] Figure 3To incorporate the robot autonomous localization method based on adaptive extended reset provided by this invention into the autonomous navigation system, the localization results at different times are visualized in RViz, a 3D visualization tool based on the Robot Operating System (ROS) framework, indicated by black arrows. This allows for an intuitive understanding of the robot's real-time and accurate localization on the map.

[0115] Example 2

[0116] An apparatus includes a memory and a processor, the memory being used to store a program that supports the processor in executing the robot autonomous localization method based on adaptive extended reset as described in Embodiment 1, the processor being configured to execute the program stored in the memory.

[0117] Example 3

[0118] A storage medium storing a computer program, which, when executed by a processor, performs the steps of the robot autonomous localization method based on adaptive extended reset in Embodiment 1.

[0119] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0120] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A robot autonomous positioning method based on adaptive extension reset, characterized in that, The method comprises the following steps: S1, installing a single-line laser radar on a differential robot, obtaining single-line laser radar data, odometer data and a two-dimensional grid map; S2, initializing a particle set, each particle containing a pose and a weight; S3, performing motion update according to the odometer data, updating the particle pose, and calculating the likelihood of the single-line laser radar data and the two-dimensional grid map; S4, calculating a new particle weight according to the likelihood; S5, comparing the size of the new particle weight and a preset threshold, if the new particle weight is less than the preset threshold, performing S6 to perform adaptive expansion reset resampling, if the new particle weight is greater than or equal to the preset threshold, performing S8 to perform regular resampling; S6, calculating an adaptive factor according to the new particle weight, and then calculating an expansion parameter according to the adaptive factor; S7, adaptively expanding and resetting and updating the weight of each particle, and performing S9; S8, performing regular resampling based on the cumulative weight and the uniform sampling interval, and performing S9; S9, calculating a weighted pose average value according to each particle pose after sampling, and obtaining a final pose estimation of the differential robot; S10, when new single-line laser radar data is received, processing the new data in a loop, and returning to perform S3.

2. The robot autonomous positioning method based on adaptive extension reset according to claim 1, characterized in that, The single-line laser radar data in S1 is obtained by a single-line laser radar installed on the differential robot, the odometer data is obtained according to the kinematics model of the differential robot, and the two-dimensional grid map is known data obtained in advance.

3. The robot autonomous positioning method based on adaptive extension reset according to claim 1, characterized in that, S2 specifically refers to: initializing the particle set. ,in, Indicates the number of particles. Indicates the first The pose of each particle is denoted as... , Indicates the first i Individual particles x Distance in direction Indicates the first i Individual particles y Distance in direction Indicates the first i The orientation angle of each particle; Indicates the first i The weight of each particle, denoted as . .

4. The robot autonomous positioning method based on adaptive extension reset according to claim 3, characterized in that, The S3 comprises: S31, motion update according to odometry data, obtaining current time the i-th particle pose i wherein, represents the distance of the current time the i-th particle in direction x, represents the distance of the current i time the i-th particle in direction y, x represents the distance of the current time the i-th particle in direction z, represents the distance of the current i time the i-th particle in direction x, y represents the distance of the current time the i-th particle in direction y, represents the distance of the current i time the i-th particle in direction z, S32, according to the single-line laser radar data and the two-dimensional grid map, calculating a likelihood of the single-line laser radar data and the two-dimensional grid map Specifically, According to the current time of day i particle pose converts single-line lidar data onto a two-dimensional grid map, calculates the coordinates of the beam of single-line lidar converted onto the two-dimensional grid map using the following logic: in, In order to be in Time of the first Single-line lidar distance measurement values, For the first Angle measurement values ​​of a single-line lidar beam. , The first The beam of single-line lidar is converted into the horizontal and vertical coordinates on a two-dimensional grid map; According to the improved likelihood calculation formula, the likelihood of the single-line laser radar data and the two-dimensional grid map is calculated by using the following logic : wherein, is represented as a coordinate the distance to the nearest obstacle, is a normalization parameter.

5. The robot autonomous positioning method based on adaptive extension reset according to claim 4, characterized in that, The new particle weights are calculated in S4 using the following logic : wherein, is the likelihood of the single-line lidar data and the two-dimensional grid map, is the number of particles.

6. The robot autonomous positioning method based on adaptive extension reset according to claim 5, characterized in that, The S6 comprises: S61, calculating an adaptation factor with the following logic: wherein, is a preset threshold value, is a new particle weight; S62, for each particle that needs to be reset , according to the adaptive factor an extension parameter is calculated; the extension parameter comprises a position extension and an angle extension , using the following logic representation: wherein, and respectively denote the minimum and maximum allowed spread radius in the environment, and respectively denote the minimum and maximum allowed angular offset.

7. The robot autonomous positioning method based on adaptive extension reset according to claim 6, characterized in that, The S7 comprises: S71, calculates the covariance matrix of the particle set at the current time point, wherein S72, calculates the weight of the i-th particle at the current time point, and uses the following logic to express the covariance matrix : w = 1 / (1 + exp(-x)) wherein, x represents the weight of the i-th particle at the current time point i ​​​ wherein, represents the distance of the i-th particle in the x-direction at the current time instant, i x represents the distance of the i-th particle in the y-direction at the current time instant, i y is the weight of the i-th particle at the current time instant, i , are the weighted average values of the particles from particle 1 to particle N in the x-direction and y-direction at the current time instant, respectively, taken as , i N , ;​​​​​​​​​​ According to Eigenvalue decomposition is performed to obtain eigenvalues With And eigenvectors With , satisfying > ; When the resampling frequency is reduced; when particles are dispersed, S72 is executed; S72, performing extended sampling on each particle that needs to be reset, obtaining each sampled particle, denoting the pose of each sampled particle as is expressed by the following logic: wherein is a direction value of the particle, represents solving the arctangent value of , , , are all particle state random offset amounts, , conform to a Gaussian distribution with a mean of 0 and a variance of , conform to a Gaussian distribution with a mean of 0 and a variance of , and step S9 is executed.

8. The robot autonomous positioning method based on adaptive extension reset according to claim 5, characterized in that, The S8 is specifically: calculating the cumulative weight of the 1st to with uniform sampling interval points using the following logic:​ in, For the first The weight of each particle, , A random number between 0 and 1 / N; for each Find satisfaction Time corresponding Once the pose of each particle after sampling is obtained, routine resampling is completed, and step S9 is executed.

9. An electronic device comprising a memory and a processor, characterized in that The memory is configured to store a program supporting the processor to execute the robot autonomous positioning method based on adaptive expansion reset according to any one of claims 1 to 8, and the processor is configured to execute the program stored in the memory.

10. A storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to perform the steps of the robot autonomous positioning method based on adaptive expansion reset according to any one of claims 1 to 8.

Images